NASA/TP-2006-214133
High Voltage Power Supply Design Guide for Space
Renate S. Bever, Arthur P. Ruitberg, Carl W. Kellenbenz, and Sandra M. Irish
July 2006
COVER CAPTION
Modular construction of the Cassini/Cassini Plasma Spectrometer (CAPS) power supply showing the combination of solid
potted, conformal coated, and bare construction. The supply has both a +16 kV and -16 kV output and both are command-
able from 0-16 kV. A +1.2 kV output floats on top of the commandable -16 kV output.
The NASA STI Program Office ... in Profile
Since its founding, NASA has been dedicated to the
advancement of aeronautics and space science. The
NASA Scientific and Technical Information (STI)
Program Office plays a key part in helping NASA
maintain this important role.
The NASA STI Program Office is operated by
Langley Research Center, the lead center for
NASA’s scientific and technical information. The
NASA STI Program Office provides access to
the NASA STI Database, the largest collection of
aeronautical and space science STI in the world.
The Program Office is also NASA’s institutional
mechanism for disseminating the results of its
research and development activities. These results
are published by NASA in the NASA STI Report
Series, which includes the following report types:
• TECHNICAL PUBLICATION. Reports of
completed research or a major significant phase
of research that present the results of NASA pro-
grams and include extensive data or theoretical
analysis. Includes compilations of significant
scientific and technical data and information
deemed to be of continuing reference value.
NASA’s counterpart of peer-reviewed formal
professional papers but has less stringent limita-
tions on manuscript length and extent of graphic
presentations.
• TECHNICAL MEMORANDUM. Scientific
and technical findings that are preliminary or of
specialized interest, e.g., quick release reports,
working papers, and bibliographies that contain
minimal annotation. Does not contain extensive
analysis.
• CONTRACTOR REPORT. Scientific and techni-
cal findings by NASA-sponsored contractors and
grantees.
• CONFERENCE PUBLICATION. Collected
papers from scientific and technical conferences,
symposia, seminars, or other meetings sponsored
or cosponsored by NASA.
• SPECIAL PUBLICATION. Scientific, technical,
or historical information from NASA programs,
projects, and mission, often concerned with sub-
jects having substantial public interest.
• TECHNICAL TRANSLATION. English-language
translations of foreign scientific and technical ma-
terial pertinent to NASA’s mission.
Specialized services that complement the STI Pro-
gram Office’s diverse offerings include creating
custom thesauri, building customized databases,
organizing and publishing research results . . . even
providing videos.
For more information about the NASA STI Program
Office, see the following:
• Access the NASA STI Program Home Page at
http://www.sti.nasa.gov/STI-homepage.html
• E-mail your question via the Internet to
help@sti.nasa.gov
• Fax your question to the NASA Access Help Desk
at (301) 621-0134
• Telephone the NASA Access Help Desk at (301)
621-0390
• Write to:
NASA Access Help Desk
NASA Center for AeroSpace Information
7121 Standard Drive
Hanover, MD 21076-1320
NASA/TP-2006-214133
High Voltage Power Supply Design Guide for Space
Renate S. Bever (Retired), Arthur P. Ruitberg (Retired), Carl W. Kellenbenz, and Sandra M. Irish
NASA Goddard Space Flight Center, Greenbelt, Maryland
National Aeronautics and
Space Administration
Goddard Space Flight Center
Greenbelt, Maryland 20771
July 2006
ACKNOWLEDGMENTS
The authors wish to honor the memory of Mr. John L. Westrom, who initiated and inspired the engineering effort at NASA’s
Goddard Space Flight Center to design and build high voltage power supplies. The authors wish to acknowledge the efforts
of, and express gratitude to, the many engineers, scientists, and technicians — especially Kenneth M. Young and Carroll
Clatterbuck — who worked in high voltage power supply design and construction, for space use. The documentation of
their experiences have helped, to a large degree, to make this High Voltage Power Supply Design Guide possible.
Appreciation is due to Dr. George Gloeckner and his team at the Space Physics Department at the University of Maryland,
for supplying information on the CHEM/AMPTE High Voltage Power Supply.
A special thanks goes to Rogenia Burton-Pendergast who typed the manuscript and put it onto a CD, with great good
spirit and much patience.
Available from:
NASA Center for AeroSpace Information
7121 Standard Drive
Hanover, MD 21076-1320
Price Code: A 17
National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Price Code: A 10
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
PREFACE
Voltage breakdown continues to be a problem in high voltage power supplies that are constructed for use in the vacuum
of space. The physics, and especially the electrical insulating properties, of gases, solids, and even liquids needs to be
understood by electronic designers in order to prevent these breakdowns.
In 1975, a document was published by NASA, namely “Spacecraft High Voltage Power Supply Construction,” by J.F.
Sutton and J.E. Stern, NASA TN D-79, April 1975. This document is now out of print and the plates for the text were
destroyed. It is with the idea of replacing and updating this former document that this present one has been written.
Some of the basic physics tables and figures have been reproduced from the former document in the first few chapters.
Obviously, the examples of power supplies are new.
The organization of this High Voltage Guide is such that each chapter has its own references at its end, before its ap-
pendices, rather than at the end of the entire book. There are more references listed as reading material than specifi-
cally referred to in a given chapter.
Throughout the book, and especially in Chapter 4, “High Voltage (HV) Parts,” the mention of manufacturers’ names
is simply a report of the part’s use by Goddard Space Flight Center and does not constitute a recommendation of one
manufacturer over another.
EDITOR’S NOTE
This document was written and compiled over the course of more than a decade with different authors and researchers
supplying data, verbiage, figures, and photographs. These and many other factors contribute to the complexity of this
document. The factors include the length of the document itself, the original media in which much of the document
components were supplied, and the different writing styles used by not only the authors of the document itself, but
also the authors of the appendices. (A number of the appendices were previously published elsewhere, but have been
included here to make it easier for readers to gain a complete understanding of the issues at hand.)
Because of these complexity issues, except for obvious typographical, grammatical, punctuation, or consistency errors,
this document is being published as submitted by the authors. For example, nonstandard ways of depicting common
symbology, such as “°K” for Kelvin instead of the standard “K,” is being allowed in this document — at the authors’
insistence — to avoid confusion with the K symbol being used elsewhere. In addition, because of the established con-
ventions in this scientific discipline, both standard and metric units are used.
As mentioned above, the appendices in this document were previously published by various authors. The text of these
appendices refers to the original figure or table numbers used in the original publication. To clarify numbering for the
readers of this document and for those who wish to reference them, however, the editor has added unique sequential
numbers in square brackets after each original number. As an example, in Chapter 3, there are two appendices labeled
“Appendix I” and “Appendix II.” Although each references a “Figure 1,” the one in Appendix I would read “Fig. 1
[3.A.I.1]”, while the one in Appendix II would read “Fig. 1 [3.A.II.1]”.
When referring to these figures in future publications, please use the unique numbering scheme (i.e., Fig. 3.A.I.1,
Fig. 3.A.II.1, etc.) in the square brackets, rather than the original numbers used in the text. Using the unique numbers
will avoid confusion and ambiguity on the part of those people wanting to look at these figures at a later date.
iii
Thank you.
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
TABLE OF CONTENTS
Chapter 1: General Information 1-1
Chapter 2: General Background on Electrical Insulation 2-1
References 2-19
Chapter 3: High Voltage Packaging for Space: Potting or Coating or Bare 3-1
References 3-11
Appendix I: Materials Proc. Doc. S-313-029: Encapsulation 3-13
Appendix II: Process Spec.: CAPS-Proc-313-1 Parylene Conformal Coating 3-37
Chapter 4: High Voltage Parts 4-1
References 4-16
Appendix I: Evaluation of High Voltage Multilayer, Ceramic Capacitors 4-18
Appendix II: Potting Process for CAPS Project High Voltage Transformers and Filters 4-29
Chapter 5: Partial Discharge or Corona Measurement 5-1
References 5-11
Appendix I: Biddle Co. Partial Discharge Detection Equipment 5-13
Appendix II: Detailed Directions for Partial Discharge Measurement 5-15
Appendix III: Directions for GLAST Project FLEX CABLE Corona Measurement 5-20
Appendix IV: MIL-PRF-49467B, May 2001. Directions for capacitors AC Corona Testing 5-23
Appendix V: AC Corona Test of Transformers at 50-100 kHz 5-24
Chapter 6: Electrostatic Field Analysis by Computer 6-1
References 6-23
Chapter 7: Design Examples of Some High Voltage Power Supplies, Mostly Since 1978, for Space 7-1
(A) -25 kV DC; HRS and FOS on the Hubble Space Telescope, by Joseph A. Gillis 7-2
(B) -15 kV DC; IMP-G, by John L. Westrom 7-8
(C) ±2500 V DC; Stepping Supply, HAPI LAPI; (also ISTP), by Henry Doong 7-10
(D) -30 kV DC; on CHEM/AMPTE, by George Gloeckner, Ph.D 7-28
(E) -700 V DC; commercial supply for BBXRT (Renate S. Bever) 7-36
(F) 3 to 5 kV DC, also 1.3 kV DC for EGRET/GRO (Arthur P. Ruitberg) 7-39
(G) 2200 V and 2900 V DC for XTE/PCA, Karen Castell- Stewart 7-48
(H) -16 kV DC and ±16 kV DC for Cassini/CAPS project (Arthur P. Ruitberg) 7-54
(I) 1300 V DC for ACD/GLAST project. 24 power supplies! (Arthur P. Ruitberg) 7-58
Authors’ Note 7-70
v
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 1: General Information
I. Use of High Voltage Power Supplies, in Space
In space, high voltage bias is needed on the following types of instruments (most are DC biases):
• Vidicon Camera Tubes (IUE, -15 kV DC)
• Digicon Detectors (HRS on HST, -23 kV DC, 1% ripple)
• Image Intensifies (FOC on HST, -26 kV DC also Ultraviolet Detector, +13 kV DC, 1% ripple)
• “Inch-Worm” piezoelectric mechanical devices (500 V DC)
• Deflection plates (1 V->10 kV steppers, 1% ripple, slew rate is ± 1 kV/ms)
• Mass Spectrometers
• Electronic Focusing Devices
• Photodetectors
• Photomultipliers (1.5-2. 5 kV)
• Channeltrons (2-»4 kV, <20 mV ripple)
• Spiratrons
• Particle Acceleration Devices, Electron Beam Ejectors
• Spark Chambers/Cloud Chambers (3-5 kV, 2% ripple)
• X-ray detectors (1-4 kV, <20 mV ripple)
• Gamma-ray detectors (2-6 kV, <50 mV ripple)
• Particle detectors in general, for protons, a and (3 rays (1-10 kV)
• Faraday Cup Modulator (150 V to 8 kV, 1% ripple, ± 10 kV/ms slew rate)
The above power supplies are for biasing purposes and usually are very high impedance supplies, capable of only
delivering currents of the order of tens of microamperes or almost no power. Some of these devices are very noise-
sensitive, e.g., x-ray detectors. The noise specifications must be supplied by the projects, e.g., no greater than 10 mV
spikes.
A second class of need for high voltage power supplies are:
• Radar Technology
• Traveling Wave Tube Amplifiers, that is for TWTAs (communications satellites such as TDRSS)
• Klystrons/Magnetrons
• Laser Technology
Except for lasers, these are also DC supplies, drawing tens of milliamperes of current. Thus power is delivered, some
of it being dissipated within the power supply, most of it in the device however. The heat generation thus becomes an
additional problem, as to how to get rid of it. Noise sensitivity here is not critical.
The Cassini/CAPS power supply of ± 16 kV was commanded ON after several years of flight to planet Saturn, and it is
performing well. Also, in orbit now are -26 kV power supplies on the Hubble Space Telescope (HST), and the -13 kV
supply on IUE has had many years of life. Amazingly, more problems occur with supplies in the range up to 1.5 kV
DC because designers do not take the high voltage problems seriously in the initial design. High voltage for space use
means anything above 200 V, which is approximately the Paschen minimum voltage for gaseous breakdown.
Chapter 1 : General Information
1-1
High Voltage Power Supply Design Guide for Space
II. Symptoms and Causes of High Voltage Failures in Space
Symptoms of failure are:
a. The performance of the power supply and the instrument is intermittent.
b. Data output from the instrument is “strange” that is, unexpected zeros and high spikes.
c. Performance degrades at high and low extremes of temperature or of high voltage or both.
d. More and more “noisiness” with time of usage.
e. The voltage and current output of the supply becomes erratic, varies wildly, and finally ends in zero output
voltages or complete breakdown.
What are the most common causes of failure?
A. Marginal Electronic Design
This means that the electronic operation of the circuit depends too sensitively on a very narrow range of parts pa-
rameters. As soon as the circuit warms up, the parts parameters change and get outside the range where the circuit
operates (e.g., HSP on ST — the high voltage transformers heated up a little and the current input rose above operating
range — then the supply shut itself off). This is an example of the least common failure mode.
B. Electrical Insulation Problems
• Defect sites in dielectric solids, that is, voids and bubbles.
• Cracks in dielectric potting or in ceramics.
• Delaminations at interfaces between different materials and around parts.
• Presence of trapped gas at intermediate pressures, that is, between about 100 torr to 10 ’ torr, that is, in
the “Paschen minimum” range.
• Impurities inside dielectric solids, that is, metal flakes, etc.
• Contaminated, flawed surfaces.
• Thermal expansion causes mechanical stresses, which is due to differential thermal expansion of differing
neighboring materials.
• Bad adhesion between neighboring materials causing delamination.
• Too high localized electric fields at sharp points on the outside of small diameter wires. Although only in
small localities, that is where trouble starts.
• Bad geometries and too small spacing on average overall, causing too high electric fields, that is
(>30 V/mil).
• Use of defective parts.
• Influx of streams of ionizing, high energy particles or radiation from cosmic bodies; or sometimes return-
ing after circular path, originally ejected from the satellite itself.
• Choice of wrong insulating materials which have too low a dielectric strength or too low a volume resistivity
or too low an arc resistance or do not machine well, that is, form microcracks during machining, or have
too high a coefficient of thermal expansion compared to the other materials in contact with them.
The above causes are the most common causes of HV failures. As can be seen, this involves the knowledge of electric
and physical properties of insulating solids and gases. Especially the understanding is important of gaseous behavior
under electric fields and at varying pressures, all the way from atmospheric (760 torr) down to very high vacuum (10 12
torr) in order to grasp the extra problems of high voltage in space, over and above the problems on the ground.
1-2
Chapter 1 : General Information
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
It is generally accepted that above 200 V DC, high voltage practices and packaging need to be used in space. If, however,
the flight path of the satellite or rocket carrying the electronics, instruments, and batteries is over the Earth’s poles,
the South Atlantic anomaly, or the “Radiation Belts” — all areas where there is a large concentration of energetic ions
and electrons — then high voltage practices need to be used from as low as 100 V on up.
Chapter 1 : General Information
1-3
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 2. General Background on Electrical Insulation
General Background on Electrical Insulation
Construction of high voltage power supplies for use in space requires an extensive knowledge of electrical insulation.
Gaseous and solid insulation are of greatest interest for space, but not insulating liquids. These are omitted here. The
behavior of gaseous discharges or gaseous breakdown in various degrees of vacuum (represented as pressures of so-
and-so many millimeters of mercury, or torr) produces the extra problems of high voltage in space. Also, the small
size and small weight imposed as a constraint on the designer of space high voltage power supplies adds to the extra
problems over and above what one would have to solve if the supply was for use on the ground and at atmospheric
pressure.
I. Gaseous Insulation
If one mounts two plane-parallel electrodes at the ends of a gas-filled tube, then the gas changes from being a good
insulator to a conducting medium when a larger and larger potential difference is applied, in volts, between the elec-
trodes. Or, one can talk about this in terms of applying higher and higher field strength, which is the potential dif-
ference divided by the length of the gas column between the electrodes, in volts/meter or in volts/mil (volts/meter =
newton/coulomb, or force per unit charge).
This progression from insulator to conductor of a gas is pictured in Figure 2.1. The voltage axis is purposefully left
blank because it depends numerically on pressure and electrode spacing and geometry. The theoretical analysis of
the ionization of the gas, the Townsend coefficients, and the avalanche breakdown can be read elsewhere (see Refer-
ences), but the useful result is Paschen’s Law which says that for any gas the breakdown voltage is a function of the
product of the gas pressure p, times the electrode spacing d. The Paschen curves for various gases can be obtained
experimentally, as in Figure 2.2, where the product pd is plotted logarithmically on the x-axis. Or in Figure 2.3, all for
air, a whole family of curves at various electrode spacings is plotted with the pressure in torr, given logarithmically on
the x-axis and breakdown voltage logarithmically on the y-axis. One can see that for a 1 cm gap of air, breakdown at
the atmospheric pressure of 760 torr would occur at about 30 kV, whereas at the pressure of 0.6 torr, it would require
only about 300 V DC or AC peak. The part of the curve where breakdown is so easy is called the Paschen minimum,
or the corona region because one can visually observe this breakdown as a colored glow (blue for air, orange-red for
neon, etc.). Another way of looking at this is to realize that at high gas pressures or gas density, the mean free path of
an electron between collisions is very short and the electron has gained too little energy to produce much ionization.
At very low gas pressures, the mean free path is good and long, and an electron can pick up much kinetic energy, but
it meets too few gas molecules to produce much of an avalanche of ions and electrons. But there is an optimum pres-
sure in between where a great many charged particle pairs are produced and breakdown is easy. That is the Paschen
minimum.
To summarize, one can see that at a product of pd about 1 torr-cm, a minimum of about 250-300 V occurs in the
Paschen curve and this, of course, varies slightly depending on gas composition, electrode shape, presence of cosmic
ray fluxes, etc. By looking at Table 1 of Earth Atmosphere data, one can see that spacecraft instruments, which must
operate while passing through altitudes of 30-65 km, are particularly prone to corona problems. But even higher up
where the gas pressure is a “safe” 10 5 or 10 6 torr or even in deep space at 10 12 torr, the pressure within a high voltage
power supply box within the spacecraft equipment compartment may well still be at 10 1 torr. The shielding box usually
has very small venting gaps for the residual air and outgassing vapors to escape into the high vacuum of space. Thus,
the pressure right at the high voltage soldering points and terminals within the box may well linger near the Paschen
minimum for several days or even weeks after launch, and turn-on of the input power to the supply should be delayed.
This is also the reason why the high voltage parts and all high voltage metal should be either solid potted or at least
Chapter 2: General Background on Electrical Insulation
2-1
High Voltage Power Supply Design Guide for Space
coated with insulating resin, and why these resins must have extremely low outgassing and low vapor pressure. See
references [26], [27], and [29],
Figure 2.1. Gaseous discharge, Voltage vs. Current Characteristic, after Refs. [1], [23]
.1 .2 .3 .5 1 2 3 5 10 20 30 50 100 200 300 500 1000
P 0 d (Torr • cm)
Figure 2.2. Typical Paschen curves for different gases; after Druyvesteyn and Penning
(courtesy American Institute of Physics), Refs. [12], [13]
2-2
Chapter 2: General Background on Electrical Insulation
V in kV, BREAKDOWN VOLTAGE, DC. OR AC. PEAK kV , BREAKDOWN VOLTAGE, DC, OR AC, PEAK
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 2.3. Paschen’s original curves; breakdown voltage in air as a function of pressure
(Iron Electrodes).
Figure 2.4. Paschen’s curve, V; field strength curve, E. (Iron Electrodes); Ref [25].
Chapter 2: General Background on Electrical Insulation
2-3
E, kV/mm, BREAKDOWN FIELD
High Voltage Power Supply Design Guide for Space
1 10 100 10 3
VOID CHARACTERISTIC DIMENSION, pm
Figure 2.5. Corona inception stress versus void size, for a cylindrical void (©1975 IEEE),
Ref. [25]. (Reprinted with permission from the Institute of Electrical and Electronics En-
gineers.)
1 10 100 10 3
VOID CHARACTERISTIC DIMENSION, (jm
Figure 2.6. Corona inception stress over the insulation versus void size, for different dielectric
constants. These curves are for a spherical void, ©1975 IEEE, Ref. [25], (Reprinted with
permission from the Institute of Electrical and Electronics Engineers.)
2-4
Chapter 2: General Background on Electrical Insulation
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 2.1. Earth Atmosphere Data, from Ref. [6]
Altitude
km above sea level
Pressure in
mm Hg or torr
Mean Free Path
ofN, molecule,
in cm
T° K
for molecules of
mean M=29
(from velocity
calculations)
0
760
6.5 x 10' 6
10
210
1.9 x 10' 5
20
42
8.6 x 10' 5
210
30
9.5
4.2 x 10' 4
235
40
2.4
1.8 x 10' 3
260
50
0.75
6.1 x 10' 3
270
60
0.21
2.1 x 10' 2
260
70
5.4 x 10' 2
6.6 x 10' 1
210
80
1.0 x 10' 2
3.2 x 10' 1
190
90
1.9 x 10' 3
2.0
210
100
4.2 x 10' 4
10.0
240
110
1.2 x 10' 4
40
270
120
3.5 x 10' 5
150
330
130
1.5 x 10' 5
400
390
140
7 x 10' 6
1000
450
150
3 x 10' 6
2500
510
160
2 x 10' 6
5000
570
As an aside, the term “corona” is often understood to encompass all forms of low current gas discharge or glow dis-
charge. In its strictest usage, corona applies to partial discharge transients or breakdown caused by high fields at one
electrode, but no current bridges the total gap between the metal electrodes all the way.
The Paschen curve can be plotted, instead of volts on they-axis, by putting field strength in kilovolts/millimeter in-
stead, such as in Figure 2.4. One can also draw derived graphs such as Figures 2.5 and 2.6. These are for breakdown
of bubbles or voids buried in a dielectric of dielectric constant K and of varying void size, to see at what externally
applied field strength in the dielectric the bubble would break down and give a transient current or corona pulse.
The reader is reminded here that an electric field equal to voltage/spacing is only numerically correct for an electric
field between plane parallel metal plates. If electrodes are spherical or cylindrical wires, then the fields are higher than
that, and in fact increase significantly the closer one gets to the electrodes. For calculations at the surfaces of spherical
or cylindrical conductors where the field is maximum, the formulae in Table 2.2 are supplied. At the surface of very
tiny diameter wires or spheres, the field strength is extremely high, and therefore, a bubble buried near a sharp point
or a tiny wire is very likely to break down.
Chapter 2: General Background on Electrical Insulation
2-5
High Voltage Power Supply Design Guide for Space
Table 2.2. Maximum Field Strength, E, with a Potential Difference, U, Between the Electrodes, for
Several Electrode Configurations (Refs. [12], [7]). Reprinted with permission by Phillips
Research Institute.
2-6
Chapter 2: General Background on Electrical Insulation
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 2.3. Minimum Sparking Constants * At Paschen Minimum Pressures, Ref. [17]
GAS
V b minimum
pd, mm Hgx cm
L/pd
volts
Air
327
0.57
1.7 x 10' 5
A
137
0.9
1.1 x 10' 5
h 2
273
1.15
1.6 x 10' 5
He
156
4.0
0.74 x 10' 5
co 2
420
0.5
1.2 x 10' 5
n 2
251
0.67
1.4 x 10' 5
n 2 o
418
0.5
1.4 x 10' 5
0 2
450
0.7
Na (vapor)
335
0.04
S0 2
457
0.33
1.4 x 10' 5
h 2 s
414
0.6
1.0 x 10' 5
* J.J. Thomson & G.P. Thomson, “Conduction of Electricity thru Gases” Vol. 2, 1933
V b = Spark Breakdown Voltage
p = Pressure
d = Gap Length
L = Kinetic Theory mean free path at 760 mm Hg.
Chapter 2: General Background on Electrical Insulation
2-7
CORONA ONSET VOLTS (rms) BREAKDOWN VOLTAGE, kV
High Voltage Power Supply Design Guide for Space
PRESSURE TIMES SPACING, torr-cm
Figure 2.7. Effect of temperature on Paschen curve, Ref. [1],
Figure 2.8. Effect of electrode geometry on breakdown characteristic in air, Ref. [1],
2-8
Chapter 2: General Background on Electrical Insulation
FREQUENCY (Hz)
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
25 50 75 100 125 150 175 200 225 250
VOLTAGE (PEAK VOLTS)
Figure 2.9. Lower voltage breakdown limit or Paschen minimum vs. frequency, (air),
Refs. [1], [30], [31].
Chapter 2: General Background on Electrical Insulation
2-9
High Voltage Power Supply Design Guide for Space
Another useful table in terms of the Paschen minimum is Table 2.3, which shows that the minimum occurs at different
voltages for different gases; for helium it is at 156 V as compared to air where it occurs at 327 V.
There are many other perturbations on the basic Paschen curves. For instance, at a particular pressure, the breakdown
voltage depends very much on what type of metal the electrodes are made of, e.g., Breakdown volts for copper elec-
trodes, 1 mm apart is 37 kV, but for steel it is 120 kV, according to Ref [9].
Effects of temperature and of electrode geometry are shown in Figures 2.7 and 2.8. Figure 2.9, from the Jet Propul-
sion Lab documents in references [30] and [31], shows that the Paschen minimum voltage depends on the frequency
of the applied voltage in the case of AC. Apparently, the higher the AC frequency above 60 Hz, the lower the Paschen
minimum.
Another very important influence on breakdown is whether a dielectric solid surface bridges the gap between the
electrodes, in other words, gas/solid interface effects. Water vapor absorbed at the dielectric surface can drastically
lower the gas breakdown near the surface. Also, slow development of conductive paths or tracks can lead to permanent
short circuiting of the high voltage. Table 2.4 is a listing of the arc resistances (in seconds) and other characteristics
of some materials commonly used in fabrication of electronic devices.
A. Pressurization and Electronegative Gases
This paragraph is quoted from Ref. [1]:
“Normally, high voltage power supplies employed on spacecraft take advantage of the high values of breakdown
voltage V b available at low pressures. It is also possible to take advantage of the high V b values at the other end of the
Paschen curve by pressurization. For example, as shown in Figure 2.10, it is possible to double V b by pressurization
to -350 kN/nr (-50 psig) with air or C0 2 or N 2 . A better approach is the use of an electronegative gas, especially SF 6 .
Molecules of SF 6 readily capture electrons and form heavy negative ions with much lower mobility than the electrons.
In addition SF 6 is stable below 423°K (150°C), is nontoxic, and does not burn. Figures 2.11 and 2.12 illustrate the im-
provement in V b with SF 6 as in [Ref 14].” A 100 kV power supply has been successfully designed using pressurization
with SF 6 for a Sounding Rocket experiment in the early 1970s, by F. Scherb, University of Iowa. He followed design
practices developed for Van de Graaf generators.”
B. Pressure Units
For ease of calculations or comparisons of graphs from different sources, the equivalences of Pressure Units are
printed below:
1 atm = 760 mm Hg = 760 torr = 1.0133 bar = 14.696 lb/in 2 (psig)
= 1.0133 x 10 6 dyn/cm 2
1 dyn/cm 2 = 10 1 N/m 2 = 10 1 Pa
1 atm = 1.0133 x 10 5 Pa
1 bar = 0.987 atm = 14.50 lb/sq inch = 1.00 x 10 6 dyn/cm 2
2-10
Chapter 2: General Background on Electrical Insulation
SPARKOVER GRADIENT IN KILOVOLTS PER INCH
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
0 100 200 300 400
PRESSURE OF AIR IN LBS. PER SQ. INCH GAUGE
Figure 2.10. Insulating strength in compressed air with stainless steel and with aluminum
electrodes producing uniform fields over Vz- and 3 /4- gaps, Ref [10]. (Courtesy of Ameri-
can Institute of Electrical Engineers.)
Chapter 2: General Background on Electrical Insulation
2-11
RATIO, SF 6 /AIR
High Voltage Power Supply Design Guide for Space
PRESSURE - ATMOSPHERES
Figure 2.11. Ratio of corona and breakdown voltage for air and SF 6 as a function of testing
conditions in nonuniform fields, Refs. [1], [14],
2-12
Chapter 2: General Background on Electrical Insulation
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
GAP SPACING (INCHES)
Figure 2.12. 60 Hz relative strength of SF 6 to dry air as a function of configuration and
spacing of electrodes. Gases at 25°C and atmospheric pressure, Refs. [1], [14],
II. Solid Insulation
This summary of solid insulation and its breakdown is divided into the following subheadings: General, Intrinsic
Breakdown, Corona Degradation or Erosion Breakdown, Thermal Breakdown, Surface Flash-Over, Thermomechani-
cal Crack Propagation, Mechanical Breakdown (other than due to thermal stresses).
A. General
One can categorize solid insulation into three classes: organic (from living organisms), inorganic materials, and
synthetic polymers.
The most useful of the above are inorganic materials and the synthetic polymers. Among the latter are the thermo-
plastics which melt in the range of 100-120°C and are flexible enough to be molded and extruded and are used for
cable insulation. Thermosetting plastics cure upon heating and develop considerable mechanical strength and hard-
ness. High voltage potting and coating compounds belong in this category (later chapter). The most obvious desirable
characteristics of solid insulation are (1) high dielectric strength in volts per millimeter (V/mm), (2) high resistivity, (3)
low power factor to reduce heating effects. Table 2.4 gives some of these electrical characteristics. The arc resistance
(measured in seconds) is also given. This is important because “slithering” surface breakdown occurs much more
often than puncture through the bulk of an insulating material. Machinability is good for epoxies, Ultem 1000 by GE
Plastics and for Vespel, but poor for Noryl, a polyurethane (EN-265).
Chapter 2: General Background on Electrical Insulation
2-13
High Voltage Power Supply Design Guide for Space
Table 2.4. Dielectric strength and arc resistance for selected insulation materials suitable for molding,
extrusion, or casting,* Ref. [1],
Material
Arc Resistance
(seconds)
Dielectric
Strength
(volts per mil)
Volume
Resistivity
S2 - cm
Dielectric
Constant
Acetal resin copolymer
240
500-2100
10 14
Acetal resin homepolymer
129-240 (burns)
500-1210
1 -6x1 0 14
Acrylic resins
no tracks
400-500
2x1 0 1 6
Acylonitril Butadiene-Styrene
71-82
310-460
10 16
Alkyd molding compound
180+
375
10 14
5. 8-6.2
Cellulose acetate
50-310
230-365
10 10 -10 14
3.5-4.0
Cellulose acetate butyrate
unknown
250-400
10 10 -10 15
3.6-6.4
Chlorinated Polyether
unknown
400
10 15
Ethyl cellulose
60-80
800
10 12 -10 15
Delrin
125-190
400
1 0 15 -1 o 16
2.7-37
Deallyl phthalates
105-140
350-400
3.9x1 0 12 -1. 8x1 0 16
6.2
Expoxies
45-300
300-550
10 12 -10 17
3.3-5. 5
Ethylene
unknown
525-550
10 8 -10 9
Fluorinated ethylene and propylene
(copolymer)
>300
500-600
>2x1 0 18
Kel-F
>360
-500-1000
2. 5-4x1 0 16
Melamine with glass fibers
180
170
10 11
Mica-glass bonded
240-300+
350-400
10 12 -10 17
Neoprene
unknown
150-600
10 11
Nylons
130-140
342-470
1 .5x1 0 1 1 —4x1 0 1 4
3.97.6
Nylons with glass fibers
92-148
400-580
3. 0-5. 5x1 0 14
Phenolic molding compound
tracks
300-400
lO^-IO 12
Phenolic molding compound with
glass fibers
0.4 to 150
100-450
Oxide Resins
unknown
500-550
10 17
Phenylene oxide resins with glass fibers
70-120
1020
10 17
Polycarbonate
120
400
2-1x1 0 16
3.1
Polychlor otrifluoroethylene
>360
530
1 .2x1 0 1 8
Polyethylene, irradiated
unknown
2500
>10 15
2.25-3.2
Polimides
230
560
10 16 -10 17
H film (5 mil)
1 83 tracks
3600
10 18
Polypropylene
unknown
750-800
>10 16
Polypropylene with glass fibers
73-77
317-475
1 .7x1 0 1 6
Polystyrene (heat resistant)
60-135
400-600
10 16 -10 17
Polysulfones
122
425
10 16
Polyetrfluoroethylene
>300
480
V
CO
Polyvinyl chloride (Flexible)
unknown
250-800
lO^-IO 14
12.0
Polyvinyl chloride (Rigid)
60-80
425-1300
>10 16
2.4
Polyvinylidene fluoride
>50
260-1280
2x1 0 14
Silicon, Mineral filled
230
390
5x1 0 1 3
4.8
Styrenes with glass fibers
28-41
354-424
3.2-37x1 0 16
Ultem 1000, GE
unknown
830
6x1 0 1 7
3.1
Urethanes
unknown
67-7.5 (60 Hz)
2x1 0 11
Viton, fluoroelastomer
unknown
500
2x1 0 1 3
Vespel
230
400
10 16 -10 17
3.0-3.5
2-14
* These values are obtained under standard test conditions and may not be obtained in engineering ap-
plications.
Chapter 2: General Background on Electrical Insulation
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
B. Intrinsic Breakdown Strength
If one were to analyze the binding energies of electrons and spacings of ions in a perfect crystalline lattice, or of a
polymer chain of an insulating solid, and calculate the electric fields necessary to tear electrons out of this lattice, one
would calculate of the order of 1 to 30 MV/cm (0.01 to 0.3 V/A) or (2,500 to 75,000 V/mil). This inherent strength is a
bulk property, and since it is dependent only on chemical composition and dielectric properties of the perfect material,
it is called its intrinsic electric strength. It is more commonly measured in the laboratory, but only under carefully
controlled conditions and especially prepared electrodes: (1) making a specimen with spherical recesses on both sides
and depositing electrodes directly onto material by evaporation, then submerging the entire specimen in a dielectric
liquid; (2) making thin slabs of the material and placing them between 1” spheres in a dielectric liquid; (3) using two
needles of a known radius of curvature at the end and known separation, and then casting resin or solid potting the
entire assembly. The dielectric fluid of immersion in (1) and (2) should have a dielectric constant greater than the solid
to be measured. The intrinsic strength can also be calculated from various quantum mechanical theoretical formulae
which appear in texts such as O’Dwyer, Ref. [16].
Practical electrical breakdown strengths found in actual insulation systems where the special precautions above are not
taken are much lower, however, than above. Practical breakdown strengths are more around 200 kV/cm (500 V/mil)
and so it must be concluded that practical insulation systems fail due to processes other than freeing electrons or ions
from the lattice by pure electrical overstress. Some of these processes are briefly presented below.
C. Corona Degradation or Chemical Erosion Breakdown
All real insulation has imperfections. These are most often gas filled cavities or voids left in the insulation during
manufacturing. The electric field, in disc shaped gaseous voids in a dielectric, is enhanced by the ratio (K dielectric/K
gas) during AC applied voltage or during ramp-up of DC voltages. The letter K here stands for dielectric constant.
Partial discharges, also called “corona discharges,” will occur across the gaseous void when its peak electric stress
equals the breakdown voltage of the gas. The partial discharges will deliver energy to the inside of the cavity wall
where they impinge and will gradually cause chemical deterioration of the wall. The problem is especially serious
with AC applied voltages where the discharges repeat every half cycle. In fact, the erosion upon AC applied voltage
and in those geometries which give rise to nonuniform fields, advances in channels that are branched. This is called
“electrical treeing.” A continuous channel finally reaches from metallic electrode to electrode causing catastrophic
breakdown. The important subject of partial discharge is treated separately in a later chapter.
Above, we have just discussed the case of an air-void in solid dielectric. The reverse is a disc-shaped solid in a gaseous
gap and the voltage distribution is again governed by the dielectric constants K of the two materials. The voltage
across the air space (where d is the thickness) is
V =V
r Air Total
J _j_ ^ A i r ^Solid
K • d
Solid Air
This works out to yield an increase in average electric stress across the air gap, and one has to be sure that it can still
withstand the full voltage at the reduced thickness after the solid disc is inserted.
Practical dielectric strength measurements show that, in volts/mil, the dielectric strength decreases with increas-
ing thickness d as roughly l/Vd, rather than staying constant. This can be explained on the basis that the larger the
thickness or volume of the dielectric, the greater the probability of flaws and bubbles and inhomogeneities. Thus, the
practical dielectric strength is lower, and this further reinforces the idea that processes other than intrinsic ones cause
the failure. Figures 2.13 and 2.14 illustrate this point.
Chapter 2: General Background on Electrical Insulation
2-15
High Voltage Power Supply Design Guide for Space
Figure 2.13. Dupont Kapton H-film corona threshold Figure 2.14. Dupont Kapton H-film dielectric strength
voltage vs. film thickness, Ref. [1], vs. film thickness, Ref. [1],
D. Thermal Breakdown
Another very important practical mechanism is thermal breakdown. DC applied voltages cause small conduction
currents to flow in a real insulator. This current flow causes localized heating created by collision of the electrons
with the lattice and are the normal I : R losses. This becomes heat, or, upon applied AC voltage, V rms , at angular fre-
quency co, the power absorbed by the insulator is (1/2 co CV 2 rms tan 3)W, where d is the phase angle between applied
voltage and the capacitative voltage drop of a C-R parallel equivalent network of the piece of insulation. This power
dissipation under AC is usually greater than upon DC applied voltage, just due to the molecular dipole oscillations,
in addition to that caused by partial discharges. The energy losses again become heat. The heating will cause general
or localized temperature increases of the dielectric. But volume resistivity and dielectric strength generally decrease
with increasing temperature as shown in Figures 2.15 and 2.16 respectively.
25 40 55 70 85 100 115 130 145 160 175
TEMPERATURE °C
Figure 2.15. Volume resistivity vs. temperature for three encapsulant materials, Ref. [1].
2-16
Chapter 2: General Background on Electrical Insulation
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
>
i"
i—
o
z
LU
DC
I—
in
o
DC
I—
o
LU
_1
LU
Q
20 50 80 110 140 170
TEMPERATURE °C
Figure 2.16. Temperature dependence of intrinsic dielectric strength, Ref. [8],
Because the heat developed cannot get away because of the generally low thermal conductivity of electrical insulation,
then this whole scenario leads to thermal runaway. The fundamental differential equation of thermal breakdown is
C
V
V • (c grad T ) = 6E 2 ,
dt
where C v = specific heat/volume, T = absolute temperature, t = time, c = thermal conductivity, 6 = electrical conduc-
tivity, and E = electrical field strength.
The 6 and c are functions of temperature as stated above. Numerical solutions show that there exists a critical field
strength F m for which the temperature of the hottest part of the dielectric asymptotically approaches some temperature
T m with time. For field strengths >F m , the temperature reaches T m in a finite time and then runs away without limit and
causes breakdown, while for field strengths <F m the temperature rises to some value and stays there without runaway
or breakdown. J.J. O’Dwyer [Ref. 16] discusses thermal breakdown in great detail.
Chapter 2: General Background on Electrical Insulation
2-17
High Voltage Power Supply Design Guide for Space
E. Surface Flash- Over
The withstand voltage of a material to flash-over along its surface is significantly lower than breakdown through its
bulk. This, combined with the fact that metallic print patterns or printed circuit trails end in sharp edges and therefore
cause high electric fields across the supporting insulator, makes surface flash-over or surface creepage or tracking
a very common mode of failure. Surface contamination and moisture contribute. Practical design stresses between
trails are limited to 12 kV/cm (30 V/mil) and must not be allowed to be higher.
F. Thermomechanical Crack Propagation
Again, cast and extruded polymers are not perfect. Minute cracks that are present due to shrinkage or due to an attempt
to confine the polymer on all sides, may grow with time once they are started due to relatively benign temperature
changes. In addition, these cracks may by chance be present in regions of high electric fields; therefore, partial dis-
charges may occur in them. Thus finally a metal to metal catastrophic failure may develop from a small crack. Bad
adhesion between cast resins and circuit components embedded therein is also a type of crack (later chapter).
G. Mechanical Breakdown (other than thermally caused)
A particular type of mechanical breakdown is failure due to mechanical overstress of the solid, produced by electri-
cally caused forces. An example is piezoelectric shear stresses which causes failure in thick discs of BaTi0 3 ceramic
capacitors of very high dielectric constants (X7R and especially Z5U formulation).
III. Liquid Insulation
The insulating liquids have not been used in High Voltage power supplies, for space flight. Thus, discussion of them
is essentially omitted other than to say that they are convenient for laboratory use. Especially in Partial Discharge
testing at atmospheric pressure, the fluorocarbon liquids, such as FC-40 and FC-43 by the 3M company are particu-
larly useful because they leave no residue on the test object immersed in them. For this reason, FC-40 and FC-43 are
superior to insulating oils.
IV. Turn-On of High Voltage in Orbit
The underlying principle for the Turn-On procedure in orbit is a calculation or intelligent guess as to how long a time
it takes for the gas pressure inside the HV power supply shielding box, and inside the HV instrument, to fall safely to
at least below 10 1 torr. In other words, although the outer space pressure may be 10 torr as soon as one is in orbit,
the pressure inside the boxes inside the spacecraft may still be hung up at 10 1 torr or so; this is because below 10 torr
or so, the molecular flow of remnant molecules through the small vent holes and gaps between boxes and their lids is
quite slow. Meantime, the materials outgas and supply more gas. References, especially to articles by Dr. J. J. Scialdone,
on this topic, are given at the end of this chapter. Some previous experiences and rules of thumb folllow:
(1) If the power supply is bare, and the voltage is very high, such as was the case on OPEN at -30 kV DC,
then waiting 4 weeks is recommended.
(2) If the power supply is only conformally coated/staked, then one should wait at least 2 weeks, prefer-
ably longer.
(3) If (a) the power supply is solid potted, and (b) if the interface to the instrument is completely insulated
against the environment, and (c) if the instrument itself is vented properly and can withstand corona,
then turn-on can be immediate. If (a), (b), and (c) are not so, then one must wait at least 2 weeks.
2-18
Chapter 2: General Background on Electrical Insulation
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
REFERENCES
1. Sutton, J.F., and J.E. Stern, “Spacecraft High Voltage Power Supply Construction,” NASA TN D-7948; April
1975.
2. Meek, J.M., and J.D. Craggs, Electrical Breakdown of Gases, Oxford, 1953, also Wiley 1978.
3. Krebs, W.H., and A.C. Reed, “Low Pressure Electrical Discharge Studies,” STL/TR-59-0000-09931, Air Force
Contract 04 (647)-309, December 1959.
4. Von Engel, A., Ionized Gases, Clarendon Press, Oxford, 1955.
5. Dunbar, W.G., “High Voltage Design Guide: Aircraft.” AFWAL-TR-82-2057 January 1983, Vol. IV.
6. U.S. Standard Atmosphere 1962, NASA, USAF, and USWV, 1962.
7. Bowers, A., and P.G. Cath, “The maximum electrical field strength for several simple electrode configurations,”
Philips Tech. Rev., 6, 1941, p. 270.
8. Alpert, D., D.A. Lee, E.M. Lyman, and E.E. Tomaschke, “Initiation of electrical breakdown in ultra high vacuum,”
Proc. Inti. Symp. Insulation of High Voltages in Vacuum, M.I.T., October 1964.
9. Anderson, H.W., “Effect of total voltage on breakdown in a vacuum,” Elec. Eng., 54, 1935, p. 1315.
10. Trump, J.G., R.W. Cloud, J.G. Mann, and E.P. Hansen, “Influence of electrodes on DC breakdown in gases at
High Pressures,” Elec. Eng., 69 , 1950.
1 1 . Burnham, J., and J.W. Williams, “High voltage insulation technology in aerospace electronic systems,” Powercon,
9 , 1982.
12. Kohl, W.H., Handbook of Materials and Techniques for Vacuum Devices. Reinhold Publishers, 1967.
13. Druyvesteyn, M.J., and F.M. Penning, “The mechanism of electrical discharges in gases.” Rev. Modern Phys.,
12 , 87, April 1940.
14. Milek, J.T., “EPIC (Electronic Properties Information Center),” AFML-RTD-AFSE, AF 33 (615)-1235, Sulfur
Hexafluoride Data Sheets DS-140, October 1964.
15. Gallagher, T.J., and A.J. Pearmain, High Voltage. Wiley and Sons, Publ., 1983.
16. O’Dwyer, J. J., The Theory of Dielectric Breakdown of Solids, Oxford Clarendon Press, 1964.
17. Cobine, J.D., Gasesous Conductors, Dover Publ, New York, 1958.
18. Howatson, A.M., Introduction to Gas Discharges, Pergamon Press, 1965.
19. Lafferty, J.M., (Ed), Vacuum Arcs, Wiley and Sons, Publ., New York, 1980.
20. Cooke, C.M., and A.H. Cookson, The Nature and Practice of Gases as Electrical Insulators, IEEE Press, EI-13,
1978, pages 239-248.
21. Papoular, R., Electrical Phemonema in Gases, American Elsevier Publ., 1965.
22. “Proc. Second Workshop Electronic Equipment at Low Air Pressure,” Tech. Memo. 33-447, JPL, March 1969.
23. Somerville, J.M., The Electric Arc, John Wiley, New York, 1959.
24. Engineering Dielectrics. Vol. IIA, Solid Dielectrics, STP 782, 1983, ASTM, Publ.
25. Parker, R.D., “Corona testing in high voltages airborne magnetics,” Proc. 1975 Power Electronics Specialists
Conf, IEEE 1975.
26. Graf, R.F., Electronic Databook, Van Nostrand Reinhold Publ.
27. Scialdone, J.J., “Internal Pressures of a Spacecraft or other system, including Outgasing Materials, in a time-
varying Pressure Environment.” Goddard Space Flight Center X-Document 327-69-534, 1969.
28. Paul, F.W., and D. Burrowbridge, “The Prevention of Electrical Breakdown in Spacecraft,” NASA SP-208, NASA/
Goddard Space Flight Center, 1969.
Chapter 2: General Background on Electrical Insulation
2-19
High Voltage Power Supply Design Guide for Space
29. Scialdone, J.J., “An estimate of outgasing of a Space Payload and its gaseous influence on the environment.” J.
S/C and Rockets, 23(4), pg. 373, July 1986.
30. Jet Propulsion Lab, Design Requirements DM 509306C, Vol I., 1978.
31. JPL Spacecraft Electronic Packaging, JPL, D8208, 1991.
2-20
Chapter 2: General Background on Electrical Insulation
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 3. High Voltage Packaging for Space: Potting or Coating or Bare
General
There are three ways of packaging high voltage assemblies in power supplies and in instruments: (1) solid potting/
encapsulating, (2) conformal coating, and (3) completely bare. Design at an early stage must take into account which
way of packaging will be used. Table 3.1 below lists some of the advantages and disadvantages of each approach.
Table 3.1. Advantages and disadvantages of packaging modes.
Mode
Advantages
Disadvantages
Examples
Solid Potting
a) Protects HV
a) Slows down
IUE tubes
Entire voltage
portion against
circuit response time.
range:
moisture and
contamination.
b) Heavy.
c) Differential
FOC tubes
b) Allows HV to be
thermal stresses
Power
energized while
between potting and
supplies for
going through
embedded parts can
FOC, HRS, HSP;
Corona Region.
cause cracks and
Almost all power
c) Electric field is
weak at surface of
potting because of
great thickness.
d) Rugged support
for parts.
e) Cushions against
vibration.
delaminations.
supplies done at
GSFC for HV.
Conformal
a) Fast circuit
a) Power supply can
HAPI LAPI by
Coating
response.
NOT function in
Doong.
Usually up to 3 kV:
b) Saves weight.
Corona region.
c) Less thermo-
b) Electric field still
ISTP Modulator
mechanical
differential stresses.
high at surface of
coating.
c) Sensitive to
contamination.
d) Long outgassing
time due to large
surface area.
supply by Ruitberg.
Bare:
a) Fast circuit
Same as a), c) and d)
CHEM by U. of
response.
immediately above.
Maryland; -30 kV,
b) Saves weight
c) NO thermo-
mechanical stresses
and delaminations.
e) Bare high voltage
metal is exposed.
Therefore, needs
long outgassing and
high vacuum.
on AMPTE.
Chapter 3: High Voltage Packaging for Space
3-1
High Voltage Power Supply Design Guide for Space
I. Potting/Encapsulation
A. Material Properties and Selection
There are a bewildering variety of resins to choose from, and a Table: “Material Properties — Electrical, Mechanical”
in W. Dunbar’s 1979 report [1], also 1988 report [14], lists an apparently enormous variety. There are certain quantifi-
able target properties for high voltage (HV) potting materials shown here in Table 3.2. This helps to narrow down
the choices, especially the “outgassing” criteria to prevent contamination of spacecraft optics. Reference documents
for “outgassing” are NASA TN D-7362 and NASA Ref. Publ. 1124. Additional polymer characteristics that need to
be considered are viscosity during the pouring of the resin, pot life, ease of handling, need for primers, adhesion to
parts, exotherm (heat generation), tear strength, crack propagation, etc. Table 3.3 from Dunbar’s 1984 [10] report
gives these considerations.
Table 3.2. Target properties for high voltage potting materials.
Electrical properties:
Dielectric constant
< 6
Dielectric strength
> 350 V/mil
Surface resistivity
> 10 12 Q
Volume resistivity
> 10 12 Q cm
Other Physical Properties
Shrinkage
< 3%
Age shrinkage
< 0.5%
Service temperature
- 55° to + 105°C
Coefficient of Thermal Expansion
< 1.5 x 10" 4 /°F
Outgasing: Total weight loss
< 1%
Condensibles
< 0.1%
Maximum cure temperature
< 100°C
Pot life
> 30 min
3-2
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 3.3. Properties of interest for insulating materials, from W.G. Dunbar, Ref. [10].
PROPERTIES OF INTEREST FOR INSULATING MATERIALS
Mechanical
Electrical
Thermal
Chemical
Miscellaneous
Properties
Properties
Properties
Properties
Properties
Tensile,
Electric
Outgasing
Resistance to
Specific gravity
comprehensive,
shearing, and
strength
Thermal
reagents
Refractive index
bending strengths
Surface
conductivity
Effect upon
breakdown
adjacent
Transparency
Elastic moduli
strength
Thermal
expansion
materials
Color
Hardness
Liability
Electrochemical
to track
Primary creep
stability
Porosity
Impact and
tearing strengths
Volume and
Plastic flow
Stability
Permeability to
surface
against aging
gases and vapors
Viscosity
resistivities
Thermal
decompostion,
and oxidation
Moisture
Extensibility
Permittivity
Spark, arc, and
flame resistances
Solubility
absorption
Flexibility
Loss tangent
Temperature
Solvent crazing
Surface
absorption of
Machinability
Insulation
coeffiecients of
water
resistance
other properties
Fatigue
Frequency
Melting point
Resistance to
fungus
Resistance to
coefficients of
abrasion
other properties
Pour point
Resistance to
aging by light
Stress crazing
Vapor pressure
Flammability
Crack propogation
Glass transition
Chapter 3: High Voltage Packaging for Space
3-3
High Voltage Power Supply Design Guide for Space
Table 3.4. Physical properties of some commonly used encapsulates, Ref. [12].
Material
Manu-
facturer
Coefficient of
Thermal Exp.
cmxlO 4
cm°C
Thermal
Conductivity
calxlO 4
cm-sec °C
Chemical
Composition
Specific
Gravity
Water
Absorption
Wgt %
Shore
Hardness
Number
Trans-
parency
and
Color
Service
Temperature
Range (°C)
Shelf
Life
Months
from
to
XR-5192
3M
Two part filled epoxy
1.53
0.36 (240 hours
at 96% R.H.l
D72
Gray
+130
12
Scotchcast 235
3M
1.6
4.0
Unfilled epoxy
1.10
1.3(1000 hour
immersion)
D55
Brown
+130
12
Scotchcast 281
3M
1.5
12.0
Two part filled epoxy
1.43
0.4(1000 hour
immersion)
D665
Brown
+155
12
Scotchcast 3
3M
2.0
4.0
Unfilled epoxy
1.10
0.8 (1000 hour
immersion)
D80
Clear
+130
12
RTV-11
GE
2.5
7.0
Silicone
1.18
A45
White
-59
+204
6
RTV-60
GE
2.1
7.4
Silicone
1.47
A60
Red
-59
+204
6
RTV-602
GE
2.9
4.1
Silicone
0.99
A15
Clear
-59
+204
6
RTV-615
GE
2.8
4.5
Silicone
1.02
A35
Clear
-59
+204
6
RTV-616
GE
2.7
6.6
Silicone
1.22
A45
Black
-59
+204
6
1090-SI
E&C*
0.54
4.1
Epoxy resin syntactic foam
0.78
0.4 (24 hour
immersion)
D78
-73
+107
6
3050
E&C
0.40
9.5
Epoxy resin
1.55
0.2 (24 hour
immersion)
D88
+125
EP-3
E&C
Two part epoxy resin
D80
Clear
-55
+120
6
IC-2
E&C
Two component urethane
A80
Clear
-55
+120
6
93-500
DOW
3.0
3.5
Silicone
1.08
<0.10(7 day
immersion)
A46
Clear
-65
+200
12
XR-63-489
DOW
3.0
3.5
Two part silicone
1.05
<1 .5 (7 day
immersion)
A35
Clear
-55
+150
12
Sylgard-182
DOW
3.0
3.5
Two part silicone
1.05
0.1 (7 day
immersion)
A40
Clear
-65
+200
12
Sylgard-184
DOW
3.0
3.5
Two part silicone
1.05
0.1 (7 day
immersion)
A35
Clear
-65
+200
6
Sylgard-186
DOW
Two part silicone
1.12
0.1 (7 day
immersion)
A32
Trans-
lucent
-65
+250
6
RTV-3140
DOW
2.9
One part silicone
1.06
0.4 (7 day
immersion)
A21
Clear
-65
+250
6
RTV-3145
DOW
4.0
One part silicone
1.12
0.4 (7 day
immersion)
A33
Gray
-65
+250
6
K230
CONAP
5.0
Two part epoxy
-1.4
0.37 (24 hour
immersion)
D65+70
Clear
12
CE-1155
CONAP
Two part solvent based
oolvurethane
Sward 70
Clear
-130
12
Epon 828 -
Versamid 140
50% -50%
Shell
General
Mills
Two part epoxy
Rockwell
M80
Solithane 113
Thiokol
Urethane prepolymer
1.07
-0.2 (24 hour
immersion)
A-35toD-60
Clear
+121
Humiseal 1B12
CTC
One part, 20% solid acrylic
1.05
0.18 (24 hour
immersion)
-59
+138
12
2# Custom
Foam 6-1104
Rogers
Foam
.083
Polyester-Polyurethane
Black
Uralane 8267
Furane
One component urethane
Clear
6
B-6-640-1
Westing-
house
Red
<12
* E&C = Emerson & Cuming
3-4
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 3.5. Electrical properties of some commonly used encapsulants, Ref. [12],
Material
Dielectric
Constant
Dissipation
Factor
Test
Frequency
Dielectric
Strength,
Volts/mil
Arc Resistance
Seconds
Surface
Resistivity
ohm
Volume
Resistivity
ohm-cm
XR-5192
4.62
3.1
100 Hz
276
168
1.5x10i 3
Scotchcast 235
5.2
0.05
100 Hz
325
1x1015
Scotchcast 281
4.9
0.05
100 Hz
375
>1x10l 4
Scotchcast 3
3.3
0.005
100 Hz
300
>1x1015
RTV-11
3.6
0.019
60 Hz
500
£100
~10 15
6.0x1014
RTV-60
3.7
0.020
60 Hz
500
O
O
A J
~10 15
1.3x1014
RTV-602
3.0
0.001
60 Hz
500
O
O
A l
~10 15
1.0x1014
RTV-615
3.0
0.001
60 Hz
500
O
O
a7
~10 15
1.0x1015
RTV-616
3.0
0.001
60 Hz
500
£100
-1015
1.0x1015
1090-SI
3.7
0.02
60 Hz
375
1x1015
3.1
0.01
1 kHz
2.9
0.01
1 MHz
3050
4.4
0.01
60 Hz
400
1x1014
4.2
0.02
1 kHz
3.9
0.04
1 MHz
EP-3
4.4
0.006
1 kHz
400
1015 /square
IC-2
5.0
0.04
60 Hz
>400
>1x10i 2
5.0
0.04
100 MHz
93-500
2.75
0.0011
100 Hz
570
6.9x10l 3
2.73
0.0013
100 kHz
XR-63-489
2.88
0.002
100 Hz
500
115
3.6x1 0 14
1x1014
2.88
0.002
10 kHz
Sylgard-182
2.70
0.001
100 Hz
550
115
2.0x1014
2.70
0.001
1 MHz
Sylgard-184
2.75
0.001
100 Hz
550
115
1.0x1014
2.75
0.001
1 MHz
Sylgard-186
3.01
0.0009
100 Hz
575
>7x1 0 1 6
2x1015
3.00
0.001
100 kHz **
RTV-3140
2.64
0.0016
100 Hz
500
50
5x10l 4
2.63
0.0006
1 MHz
RTV-3145
2.81
0.0015
100 Hz
600
50
5.0x1014
2.78
0.0028
1 MHz
K230
3.35
0.03
1 MHz
2000 (5 mil film)
1 .25x1 0 14
IxIOl 4
CE-1155
3.50
0.0142
100 Hz
3000 (2 mil film)
5.66x1 0 14
1.18x101®
3.43
0.0138
1 kHz
1045 (22 mil film)
Epon 828 -
3.23
0.0036
60 Hz
5.5x1015
1.22x1015
Versamid 140
3.19
0.0070
1 kHz
50% - 50%
2.99
0.019
1 MHz
Solithane 113
2.8-5.0
0.014-0.162
1 kHz @ 80° F
340-512
1.5x1015
7x10l 2 t0
4. 5-5.1
0.006-0.079
1 kHz @ 185°F
3.6x1014
Humiseal 1B12
2.8
0.01
1 MHz
6000V
2.5x10l 4
2# Custom Foam
97% voids inter-
connecting calls
(MIL-I-46058B)
Uralane 8267
4.4
0.049
1 kHz
2500 (3 mil film)
149
3.0x1012
3.6
0.053
1 MHz
B-6-640-1
+
1200 (5 mil film)
126
2x1 0 13
Chapter 3: High Voltage Packaging for Space
3-5
High Voltage Power Supply Design Guide for Space
Table 3.6. Mechanical properties of some commonly used encapsulates, Ref. [12].
Material
Tensil
Strength
(kpsi)
Tensil
Elongation
%
Pot Life,
hours
Viscosity,
Poises
Principal Characteristics
XR-5192
0.995
75
High arc and track resistance
Scotchcast 235
1.3
75
0.25
15
Low viscosity, permanent flexibility
Scotchcast 281
2.1
45
0.3
480
Permanent flexibility, high temp, stability
Scotchcast 3
4.4
1.8
0.3
16
Lowest viscosity, excellent electrical properties
RTV-11
0.35
180
1-6
120
Flexible
RTV-60
0.80
130
1-5
500
Flexible, high temperature
RTV-602
0.10
200
0.5-8
12
Transparent
RTV-615
0.925
150
~4
40
Transparent, high temperature
RTV-616
0.925
125
~4
90
High temperature
1090-SI
18
Low density
3050
5
Low viscosity
EP-3
6
2.4
Surface coating, good mechanical and water resistance
IC-2
400
4.0
Surface coating, good mechanical and water resistance, high temperature
93-500
0.790
110
1
80
Low weight loss in hard vacuum
XR-63-489
0.90
100
8
50
Transparent, flexible, for laminating glass
Sylgard-1 82
0.90
100
8
30
Low viscosity, low cure shrinkage, wide temperature range
Sylgard-1 84
0.90
100
2
30
Low viscosity, low cure shrinkage, wide temperature range
Sylgard-1 86
0.70
420
2
450
High strength, wide temperatue range
RTV-3140
0.30
350
350
Clear conformal coating, no acetic acid evolved during cure
RTV-3145
0.70
675
Clear, high strength, non-corrosive, wide temperature range
K230
2.0
1-1.5
Clear epoxy, kit form
CE-1155
6
.72
Coating with good moisture and abrasion resistance
Epon 828 -
Versamid 140
50% - 50%
8.3
8.3
-160
Solithane 113
0.16-3.2
60-120
0.3-8
200
Versatile, soft to extremely rigid depending on catalyst used
Humiseal 1B12
2# Custom
Foam 6-1104
1.05
0.3 strokes
Low viscosity coating
Excellent vibration and shock protection
Uralane 8267
1. 5-3.0
Repairable, solder-through, transparent coating
B-6-640-1
12
<12
Tough, resilent nontracking surface coating
3-6
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 3.7. Potting and coating compounds used most often at GSFC.
Potting
Resin:
Primer:
Volume
Resistivity
in Q cm:
25°C
Dielectric
Constant:
25°C
Coeff. of
Thermal
Expansion
per °C
Glass
Trans.
Temp. Tg
Dielectric
Strength
V/mil
DC 93-500
DC 36060
6.9 x 10 13
(6.2 x 10 14 )
2.7 at
0.1 MHz
300 x 10' 6
-115°C
550
Uralane
5753 LV
PR-1
1.2 x 10 16
(2.3 x 10 16 )
2.9 at
1 MHz
200 x 10' 6
-60°C
450-500
Conathane
EN-11*
PR4-20 or
Epon828/
Versamid
140,50%-
50%
4.3 x 10 15
at 25°C; but
4.8 x 10 11 at
130°C
2.9 at
1 MHz
140 x 10' 6
-70°C
450-500
Devolatil-
ized
RTV615
SS 4155
4.5 x 10 13
(1 x 10 15 )
3.0 at 1 KHz
& 100 Hz
270 x 10' 6
-120°C
550
2B74
Polyure-
thane
1 x 10 15
2.9 at
1 MHz, 4.2
at 1 00 Hz
100 x 10' 6
450
Hysol
PR 18M
2 x 10 13
3, at 1000
MHz
Parylene C
7 x 10 16
3, at 1 MHz
3 0x1 O' 6
-5000
V/mil: 1 mil;
-550 V/mil
at 1/8 inch.
* EN-11A resin has been discontinued; instead, use EN-4A resin with EN-11B catalyst.
Chapter 3; High Voltage Packaging for Space
3-7
High Voltage Power Supply Design Guide for Space
Table 3.8. Some glass transition temperature (T g ) measurements; excerpted from S.Y.
Lee, Ref. [7],
MATERIAL
Tg, °C, by TMA
Tg, °C, by DMA
Lexan
146, 148
155
Plexiglass (regular)
110, 111
119
ULTEM 1000
215
Epon 828/Versamid 140
51%, 49%
42, 43
71
Conap EN-11
-70, -68
-39
Uralane 5753LV
-59, -62
Solithane 113/113-300
Compound #1
-7, -7
25
Humiseal-2B74
100/85
5, 5
37
Note: The first three materials are thermoplastics. The others are thermosetting materials cured at room temperature
for 7 days, except for Epon 828/V140, which was cured at 70°C for 3 h and then at 80°C for 1 h.
Unfortunately, there is no single , ideal potting material that has all the desirable properties. Choice of the potting ma-
terial is usually made from among (1) silicone rubbers or RTVs, and (2) polyurethanes. The encapsulant properties
from the earlier GSFC document by Sutton and Stern [12] are repeated here for convenience as Tables 3.4, 3.5, and
3.6, with the addition of the most recently used potting materials at GSFC, namely Uralane 5753 LV and Conathane
EN-11 as in Table 3.7. Epoxies, type (3), used for small jobs such as parts encapsulation, often are Epon 828/Versamid
140 or Stycast 3050.
A few words need to be said about the Glass Transition Temperature, T g in the last column of Table 3.7. T g is a polymer
material’s transition temperature below which the “plastic” chemical is hard and brittle and above which it is rubbery
and softer. Actually the transition occurs over a small range of temperature of about 20°C. Also, in the literature,
from different investigators, the T g can be different by 20-30°, depending on the method of measurement, whether it
is by Thermomechanical Analysis (TMA) or Dynamic Mechanical Analysis (DMA). The single temperature given in
Table 3.7 is a rounded average from various sources. It is important that the lowest expected temperature during the
flight and the lowest test temperature during the thermal vacuum testing is well above the T g of the chosen potting or
coating material. Table 3.8 gives some of the data from the research and article by Shen Yen Lee, Ref. [7],
Mention must also be made of plastic materials that are used below their glass transition temperatures. Two good ex-
amples of these are the older VESPEL, a polyimide, and the newer Ultem 1000, a polyetherimide by General Electric
Plastics. These latter materials have excellent machinability and are used as housings for transformer molds, diode
housings, etc. This is seen in Chapter 7 in the Cassini/CAPS high voltage power supply. There, the Ultem 1000 was
chosen to house the parts in the high voltage multiplier stack and also for the transformer molds. Compared to other
machinable thermoplastics, such as polycarbonate, the Ultem 1000 is better in that it does not craze upon machining,
and there is a definite procedure for annealing afterwards to alleviate stresses. Ultem 1000 also has good adhesive
3-8
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
properties to the materials that might be used for potting or coating after the insertion of the electronic parts, such
as to EN-11 or to Parylene. Furthermore, the Ultem fulfills the low outgassing requirements. Its glass transition
temperature is 215°C.
B. Potting Process
The manufacturer’s recommendations for the processing of the selected potting material should be carefully followed.
In addition, very important steps to ensure successful potting for HV use in space must be taken. The Materials
Processing Document, “Encapsulation of HV Power Supplies using Urethane Resin,” No. S-313-029, Feb. 1993 is
reprinted in this report as Appendix I in this chapter (Chapter 3).
1) To ensure good adhesion of the encapsulant to all the parts and cables of the HV assembly, this assembly must
be scrupulously cleaned. Ultrasonic cleaning can be done with Freon TMS or vapor degreasing with Freon or
Trichloroethylene; or a 50:50 mix of Freon and 200 proof ethyl alcohol can be used. Materials experts should be
consulted.
Certain outer surfaces must be pre-etched just before potting. Examples are Teflon, Delrin, and Diallyl Phthalate.
Again, a materials expert should recommend the particular etchant solution.
Priming should then be done on the parts, the board, and the bus wires. Priming is sometimes done selectively,
such as omitting primer on the interior of the box walls except where the HV cable penetrates, or omitting it on
coated capacitors. Each primer selected for a later encapsulant must, of course, have a very high resistivity and
must be applied very, very thin.
2) To avoid gas entrapment or bubbles. After the resin, hardener and catalyst of the potting compound have been
mixed, the mixture must be pre-degassed in a very large cup under a crude vacuum of about 1-5 torr. This “poor”
vacuum is desirable because it will release the stirred-in air bubbles, but will not boil off the hardener. When froth-
ing ceases, the vacuum is released, and the mixture transferred to a smaller glass beaker, and it and the assembly
to be potted placed in a “vacuum-pouring” system. This is pumped down to about 1 torr and pumped for about
20 min. Then, without opening it to the atmosphere, the glass beaker is tipped by manipulation from outside the
vacuum system, and the resin is poured into the electronic assembly. There should be very little or no air bubbling
out from under the resin now. If the vacuum is too “good,” there might be bubbling or bumping due to boiling of
the hardener. Finally, the pump is turned off and the system brought up to atmospheric pressure by admitting air or
dry nitrogen. This forces the resin into small spaces in the electronics providing the bottom of the box containing
the electronics is leak-tight.
The curing is now carried out at the proper temperature and atmospheric pressure (never partial pressure). Over-
pressure of several atmospheres has been used on transformers, but care must be taken that this does not prevent
bubbles from breaking through the surface of the resin.
3) To prevent tearing loose or cracking of the encapsulant. Cure shrinkage and much larger coefficients of thermal ex-
pansion of potting resins than the inorganic embedded parts, require design and processing attention. Even though
the potting polymers are plastics, nevertheless they demand to be given a definite volume at a given temperature.
If an attempt is made to constrain the potted volume with the rigid metal walls of a shielding box on all sides and
force adhesion on all sides, then during temperature changes, the stresses in the constrained potting material will
cause cracking in the potting material or tearing loose from components and parts. These cracks or thin gaps of
poor adhesion quickly fill with vapor, and will have partial discharges in them if they are in a high electric field
region (E) leading to noise, degradation, and finally total breakdown after time.
For the above reasons, the box walls, where the E is low, should not be primed, except at cable penetrations, permit-
ting tearing loose of the potting as a thermomechanical stress relief. Also, one of the six surfaces of a rectangular
block potting job should be left free. The free surface, usually the top of the potting, can be painted with a mix of
the same resin loaded with carbon-black to make it sufficiently conductive to bleed off charges to the grounded box
wall. Alternatively, if necessary, the potting can be done into a totally removable Teflon mold, all the free surfaces
Chapter 3: High Voltage Packaging for Space
3-9
High Voltage Power Supply Design Guide for Space
later to be painted with a carbon containing resin for bleed-off of electrostatic charge. The free potted HV circuit
block can then be mounted into a fairly snug-fitting (but not too tight) shielding box and the coating connected to
the wall to bleed off electrostatic charge. Obviously, this mechanical mounting design needs to be planned well
ahead.
II. Conformal Coating
As stated earlier, conformal coating and staking rather than solid potting may be preferred up to about 3000 V. A
materials expert should be consulted as to the best processing for a given choice of coating resin. For instance, if EN-
1 1 is used for coating, it is brushed on without dilution (being sure the brush loses no hairs) after initial degassing
of the resin in vacuum. The coated board is then immediately degassed again in vacuum, and then, at atmospheric
pressure, left to cure at 50°C for 24 h in a horizontal position.
Alternatively, the literature recommends thinning out of the coating resin with volatile thinner and applying three
successive coatings, by spraying at right angles to each other. It is obvious that the circuit boards, just as for potting,
have to be cleaned and perhaps even primed initially, as per Ref. [15],
The high voltage transformer with its hundreds of turns of very thin (AWG 32 and up) magnet wire presents special
problems. It must be wound relatively loose, in a basket weave, and be pre-impregnated with a low viscosity resin
(epoxy), without trapped air in the wire windings. This is especially important if the HV circuit assembly is later only
conformally coated. In that case, if there is trapped air in the transformer it will diffuse out through the thin coating
in a few weeks or months into the vacuum of space, leaving the inner HV windings surrounded by air gaps at partial
pressures. These will then have many partial discharges due to the alternating high voltages at high frequencies, lead-
ing to erosion and eventual catastrophic breakdown between turns.
More recently, successful conformal coatings have been applied with Parylene-C to high voltage circuit boards for
photomultiplier power supplies, to 3 kV. Special board-mounting techniques of ceramic multilayer capacitors must be
followed, and this involves leaving at least 0.005 in of spacing between the bodies of the ceramic capacitors and the top
surface of the circuit board. The coating must be extremely thin, that is 0.001 in, so that the ceramic chip capacitors
are not rigidly attached by their bodies to the board. This Parylene-C coating must be done with special techniques
and special equipment for vapor deposition in vacuum. A Goddard Space Flight Center Process Specification for coat-
ing of Cassini/CAPS-PROC-313-1 is attached as Appendix II to this chapter. This is a much higher voltage supply of
16 kV and is of modular construction. That is, only some portions are Parylene coated.
III. Completely Bare High Voltage Power Supplies
This has been used successfully by the University of Maryland Space Physics Department, College Park, Maryland,
for a power supply at -30 kV for the AMPTE project. Many special techniques in construction must be used for such
an unprotected supply. A person to consult is Dr. G. Gloeckner, Head of the Space Physics Department there. Some
of the most obvious techniques are:
• Sufficiently large volume allowed
• Corona rings or field “Smoothers”
• Especially designed HV feedthrough from shielding box
• Absolutely no plastic or polymer coatings on parts such as cables, capacitors, resistors, etc., so as to
achieve low vapor pressure
• Large vent hole in shielding box; vent hole modified with metal screen
• All testing done in high vacuum
3-10
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
• Three weeks of outgassing allowed for HV supply, in shielding box, in high vacuum, before voltage
turn-on; this is especially important in orbit
• One of the HV transformers finally had to be Parylene-coated, after all.
See Chapter 7 for more detail.
REFERENCES
1. Dunbar, W.G., and RA. Tjelle, “Manufacturing Technology for Airborne High Voltage Power Supplies,” Vol.I,
Tech Kept. AFML-TR-79-4018, Vol. I, Boeing Aerospace Co., AFML, U.S. Airforce, Feb. 1979.
2. Tweedie, A.T., and A. Ismail Abdel-Latif, “Exploratory Development of Space Qualified Potting Compounds,”
G.E. Space Div., Contri.#F33615-82C-5007, AFML, U.S. Airforce.
3. Hudgins, W.P., B.G. Raplee, and C.J. Tirma, “Space Qualified Potting Compounds,” TRW Co., AFSC, AFML
Contr. #F33615-79-C-5098, U.S. Airforce.
4. Dunbar, W.G., “High voltage power supply materials evaluation,” IEEE 1982 Inter. Symp. Elect. Insulation, June
1982, p. 46.
5. Dunbar, W.G., “High Voltage Design Guide: Aircraft,” Vol. IV, Boeing Aerospace Co., AFWAL-TR-82-2057, Vol.
IV, AFWAL, U.S. Airforce, Jan. 1983.
6. Ibid, “High Voltage Design Guide: Spacecraft,” Vol. V, Jan. 1983.
7. Lee, Sheng Yen, “Thermomechanical Properties of Polymeric Materials and Related Stresses,” SAMPE Quarterly,
21(4), July 1990.
8. Dakin, T.W., “Partial discharges with D.C. and Transient High Voltages,” Proc. Nat. Aerospace Electronics Conf,
Dayton, Ohio, May 1978.
9. Cambell, W.A., Jr., R.S. Marriott, and J.J. Park, “Outgassing Data for Selecting Spacecraft Materials,” NASA Ref.
Pub. 1124, 1984.
10. Dunbar, W.G., “Manufacturing Technology for Improved High Voltage Power Supply Packaging,” Boeing Aero-
space Co., AFWAL-TR-83-4143, May 1984.
11. Bunker, E.R., J.R. Arnett, K.H. Li, and J.L. Williams, “Manufacturing Methods and Technology for Electro-
magnetic Components, Vol. I and II,” Hughes Aircraft Co., Report No. FR-80-76-1254R, HAC Ref. No. D8712,
Dec. 1980.
12. Sutton, J.F., and J.E. Stern, “Spacecraft High Voltage Power Supply Construction,” NASA Tech. Note D-7948,
April 1975.
13. Graf, R.F., Electronic Databook. Van Nostrand/Reinhold Publ.
14. Dunbar, W.G., “Design Guide: Designing and Building HV Power Supplies,” AFWAL-TR-88-4143, Vol. II,
1988.
15. GSFC, “Conformal Coating of HV Printed Wiring Boards with a Room Temp. Curing Urethane Resin,” Materials
Processing Document, S-313-019, Rev. A, Sept 1992.
16. GSFC, “Encapsulation of HV Power Supplies using Urethane Resin,” Materials Processing Document, S-313-029,
Feb. 1993.
17. GSFC, “Parylene Conformal Coating of the CASSINI/CAPS High Voltage Power Supply,” Materials Proc. Doc.
CAPS-PROS-313-1.
Chapter 3: High Voltage Packaging for Space
3-11
High Voltage Power Supply Design Guide for Space
APPENDICES to CHAPTER 3
Appendix I: Materials Proc. Document S-313-029
Encapsulation of High Voltage Power Supplies, using a Polyurethane Resin. February 1993. The great detail herein
is thought to be of help to technicians, having to actually do the potting. There is also Proc. Doc. S-313-019 on Con-
formal Coating with Polyurethane.
Appendix II: Process Specification CAPS-PROC-313-1, Rev.
Parylene Conformal Coating of the Cassini/CAPS HV Power Supply. The unusual vapor deposition in vacuum for
conformal coating is described in detail.
3-12
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Appendix I to Chapter 3
MATERIALS PROCESSING DOCUMENT
S-313-029
Encapsulation of High Voltage
Power Supplies Using a Polyurethane Resin
February 1993
Materials Branch
Office of Flight Assurance
NASA/Goddard Space Flight Center, Greenbelt, MD 20771
Chapter 3: High Voltage Packaging for Space
3-13
High Voltage Power Supply Design Guide for Space
Encapsulation of High Voltage
Power Supplies Using a Polyurethane Resin
SECTION
HEADING
1.0
Overview
2.0
Purpose
3.0
Scope
4.0
Approval Requirements
5.0
Quality Assurance Requirements
5.1
Workmanship Requirements
5.2
Power Supply Final Acceptance Criteria
6.0
Space Flight Materials Control and Use Requirements
7.0
Material Supply and Source List
8.0
Processing Facility and Essential Equipment Requirements
9.0
Power Supply Cleaning Process Requirements
10.0
Priming Requirements
11.0
Encapsulating Resin Application Requirements
12.0
Encapsulating Resin Curing Requirements
13.0
Accessories Cleaning Requirements
14.0
Applicable Documents
15.0
Abbreviations
16.0
Figures and Tables
17.0
Attachments
3-14 Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
LIST OF FIGURES
Figure Title
1 An Assembled High Voltage Power Supply without Resin Encapsulation
2 A Customized System Used to Encapsulate High Voltage Power Supplies in Vacuum
3 A Vacuum System with a Turn Table and Rotating Resin Pouring Container
4 An Illustration of the Encapsulation Resin Being Poured under Vacuum
5 An Assembled Power Supply Encapsulated with a Polyurethane Resin
6 Soxhlet Extractor Assembly-Two Units
Attachment I Standard Material Certification (SMC)-EXAMPLE I
Attachment II Work Request II
Attachment III Mixing Record III
LIST OF TABLES
Table Title
1 Properties Reference of the Polyurethane Encapsulant Resin and the Epoxy Primer
Chapter 3: High Voltage Packaging for Space
3-15
High Voltage Power Supply Design Guide for Space
1.0 OVERVIEW
An insulating, encapsulating, polyurethane resin is applied to the cavity area of a high voltage power supply for sev-
eral reasons. First, the encapsulat'd material must electrically insulate electrical parts and critical surfaces. Second,
it must provide maximum moisture protection and environmental isolation. Third, it must have ultimate adhesion to
the electrical parts and surfaces for the duration of the space flight mission.
This document defines approved procedures for encapsulating high voltage power supplies used in space flight ap-
plications. It specifies, in detail, all essential equipment, necessary materials, cleaning, masking, priming, end item
acceptance criteria and approval requirements.
Personnel involved with encapsulating high voltage power supplies according to the procedure in this document are
required to be trained and qualified in all work aspects described by the document, including the precautions and
controls associated with handling polymer materials.
2.0 PURPOSE
The procedures outlined in this document are intended to produce uniform, repeatable and consistent results and high
quality space-flight approved, encapsulated high voltage power supplies.
In the high voltage encapsulating process, the encapsulating material is applied to the high voltage cavity area under
vacuum to minimize entrapped air bubbles and moisture. The prime objective of the overall process is the production
of a finished encapsulated high voltage power supply containing a polyurethane resin with no entrapped air bubbles,
good adhesion and electrical isolation.
3.0 SCOPE
This document contains the Goddard Space Flight Center (GSFC) approved procedure for encapsulating high voltage
power supplies used on space flight missions.
This document is available for general use by GSFC or other NASA projects or by contractors, subcontractors,
universities or other individuals engaged in building space flight hardware, but its use is subject to the provisions in
Section 4.0.
4.0 APPROVAL REQUIREMENTS
Prior approval by the GSFC Materials Assurance Office (GSFC Code 313.A) or the personnel listed on the docu-
ment sign-off page and the project for which the work is performed, shall be required before use of this document is
permitted.
Any polymer or other material substitutes (materials not specified herein including solvents, wipes, etc.) used to
encapsulate high voltage power supplies shall require prior approval by GSFC Code 313. A before their use in a space
flight application is allowed.
5.0 QUALITY ASSURANCE REQUIREMENTS
5.1 WORKMANSHIP REQUIREMENTS
Space flight quality, encapsulated, high voltage power supplies are the result of careful workmanship by qualified
personnel performing the work. Personnel involved in the encapsulation of power supplies for space flight hardware
according to the procedures in this document shall be trained and certified as qualified and competent in all work
3-16
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
aspects encompassed by the document. Certification and training of personnel performing the project work shall be
considered the responsibility of the Quality Assurance section of the organization performing the work.
Upon request, GSFC Code 313.2 personnel may be available for assistance during the encapsulating process. It is
strongly recommended that personnel attend the space flight approved polymer training course given at NASA Jet
Propulsion Laboratory as part of the certification process.
5.2 POWER SUPPLY FINAL ACCEPTANCE CRITERIA
All visual inspections shall be performed with a lighted magnifier having 10X magnification.
Measured Shore A hardness values shall be recorded in the appropriate documentation associated with the flight
hardware.
As a precaution to protect flight hardware from electrostatic discharge (ESD), the completed, encapsulated power
supplies should be stored or moved in a clean ESD approved bag or other approved container.
The encapsulating resin and its hardness test sample shall meet the following requirements before the unit is
acceptable for space flight use:
a. The encapsulating resin hardness test sample shall be in a fully cured state, tack free to the touch of a plastic
probe and shall be tested for Shore hardness. After testing, the hardness test sample shall exhibit a Shore A
hardness within ±5 units of the value measured at the time of material receipt.
b. The encapsulated assembly shall be in a fully cured state and tack free to the touch of a plastic probe.
c. There shall be no cracks, cavities, blisters, tears, bums or discoloration on or in the assembly encapsulant
material except as allowed herein.
d. The encapsulated assembly shall have no separations or delaminations on the critical surfaces or electrical
parts. Critical surfaces shall be those surfaces where electrical parts or cable wires are attached to the high
voltage assembly, including the part’s surface. Separation of the resin from the container side walls is ac-
ceptable.
e. There shall be no entrapped air bubbles of any size in any direction in the critical area of the encapsuled
assembly. The critical area shall be defined as the entire volume of the encapsulated assembly except near
penetration of the high voltage output cable.
f. There shall be no exposed electrical parts or wire insulation as a result of incomplete encapsulant material
coverage.
g. The encapsulated assembly shall be free of any particulate contamination such as dirt, hair, dust, etc.
h. Critical surfaces and electrical parts shall be primed for maximum adhesive strength as described herein
(see Section 10.0).
i. There shall be no encapsulant material or primer residue on any of the non-encapsulated surfaces after
completion of the encapsulation work.
j. The encapsulant surface shall exhibit no cuts or scratches due to post encapsulation resin trimming.
6.0 SPACE FLIGHT MATERIALS CONTROL AND USE REQUIREMENTS
There are specific control requirements for space flight approved polymer materials and their use in space flight ap-
plications. The requirements shall be follows:
a. Upon receipt, the polymer materials shall be verified against the appropriate documentation to ensure that
the correct materials were delivered. A Standard Materials Certification (SMC) or Certificate of Compliance
Chapter 3: High Voltage Packaging for Space
3-17
High Voltage Power Supply Design Guide for Space
(CofC) form from the manufacturer shall accompany the materials and shall include lot number, cured and
uncured properties, date shipped, date manufactured and the manufacturer’s expiration date. See Attachment
I.
b. The SMC or CofC and the purchase order shall be filed for record keeping, material property test verifica-
tion, and traceability of the material.
c. An inspection label shall be attached to each material container providing information on the material type,
date of receipt, date of manufacture, manufacturer’s expiration date, lot number, storage requirements, and
Shore hardness of the cured resin. A Shore hardness test shall be performed by the receiving laboratory upon
receipt of the materials and shall be completed before any high voltage encapsulating work of flight hardware
begins. The Shore hardness measured upon receipt of the materials shall be verified against the SMC or
CofC test values and recorded on the materials inspection label along with the test date. The material shall
not be used if it does not exhibit proper cure and Shore hardness of ±5 units of the manufacturer’s specified
value.
d. All space flight materials shall be properly stored in a separate cabinet, refrigerator or freezer as required
by the temperature specifications in the SMC or CofC. This storage cabinet shall be reserved only for space
flight materials, and shall be labeled “FOR SPACE FLIGF1T USE ONLY,” which have not exceeded their
expiration date. Material expiration dates shall be inspected weekly. Expired materials shall be removed from
the space flight material storage cabinet and stored in a separate cabinet. Expired materials shall be labeled
“NOT FOR SPACE FLIGF1T USE” and disposed of properly.
e. Materials shall not be used for space flight applications after they have exceeded their manufacturer’s expira-
tion date unless an approval for an expiration date extension has been granted. An extension of the materials
expiration date shall require the approval of GSFC Code 313. A personnel. An extension is generally granted
for 30 days. The request for extension shall be submitted in writing 3 weeks prior to the materials expira-
tion date. Approval shall be granted only after the soon-to-expire material has been tested at GSFC and has
demonstrated proper uncured (visual inspection — color, lack of crystallization, etc.) and cured properties.
Any material exceeding the manufacturer’s expiration date by more than 90 days shall not be used.
f. Shelf life may be extended by the contractor, providing that the contractor submits a procedure for shelf
life extension approval. This procedure must be submitted prior to the material’s use and approved by the
Materials Assurance Office, Code 313.A.
g. The preparation and application of polymers for space flight use shall be documented and controlled using
a work request and mixing record form. The work request and mixing record shall be fully completed and
filed for record keeping. An acceptable example of a work request and mixing record form can be found in
Attachments II and III, respectively, of this document.
h. As required by law, all materials used in this document should have the appropriate Material Safety Data Sheet
(MSDS) on file and readily available as needed.
3-18
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
7.0 MATERIAL SUPPLY AND SOURCE LIST
The following is a list of materials used in this procedure or manufacturers of these items.
Materials
Source
Acetone, A- 18
Fisher Scientific Co.
Pittsburg, PA
Aluminum Sheet
2.5 x 2.5 x 0.015 inches thick
(63.5 x 63.5 x 0.38 mm)
Local Source
Bag, Plastic
Clean Room Products, Inc.
Aclar LB-522 (22C)
Ronkonkoma, NY
Bag (Plastic), ESD
3M Austin Center
3M 2100 Series
Austin, TX
Beaker, Glass lOOcc
Fisher Scientific Co.
Pittsburgh, PA
Beaker (Plastic),
Airlite Products
Polyethylene 16 oz (454 g)
Omaha, NE
Brush, Gordon
Gordon Brush Manufacturing Co.
AL 25IN
Los Angeles, CA
Brush, Sable,
1/6, 1/8 and 3/8 inch diameter
(1.6, 3.1 and 9.5 mm)
Local Source
Conathane EN-ll-A*
Conap, Inc.
Olean, NY
Conathane EN-ll-B
Conap, Inc.
Olean, NY
Dryer, Forced Air 09-201-5
Fisher Scientific Co.
Pittsburg, PA
Epon 815 Epoxy
E.V. Roberts
Culver City, CA
Ethyl Alcohol, 200 Proof
Midway Grain Products
Perkin, IL
*EN-11A is no longer available; use EN-4A with EN-11B.
Chapter 3: High Voltage Packaging for Space
3-19
High Voltage Power Supply Design Guide for Space
7.0 MATERIAL SUPPLY AND SOURCE LIST
The following is a list of materials used in this procedure or manufacturers of these items
Materials
Source
Face Mask with Organic Filter 17-674
Fisher Scientific Co.
Pittsburgh, PA
Gauge, Thickness, Wet Film
Model 115, GR660
Pacific Scientific
Silver Spring, MD
Gloves, Polyethylene, 11-394
Fisher Scientific Co.
Pittsburgh, PA
Glass Beads, 11-311C
Fisher Scientific Co.
Pittsburgh, PA
Heptane
03008
Fisher Scientific Co.
Pittsburgh, PA
Isopropyl Alcohol
A-417
Fisher Scientific Co.
Pittsburgh, PA
Primer
PR-420— A
Products Research & Chemical Corp.
Gloucester City, NJ
Primer
PR-420— B
Products Research & Chemical Corp.
Gloucester City, NJ
Spatula, Plastic
Part No. 6169-0010
Nalge Co.
Rochester, NY
Spatula, Stainless Steel
14-375-10
Fisher Scientific Co.
Pittsburgh, PA
Swabs, Cotton
Local Source
Swabs, Foam
TX-700B
Texwipe Co.
Hillsdale, NJ
Tape, Kapton*
K-102
Connecticut Hard Rubber Co.
New Haven, CT
Tape, Teflon*
HM 430
Connecticut Hard Rubber Co.
New Haven, CT
3-20
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
7.0 MATERIAL SUPPLY AND SOURCE LIST
The following is a list of materials used in this procedure or manufacturers of these items.
Materials
Source
Versamid 140
E.V. Roberts
Culver City, CA
Wipes, Cotton
Texwipe Co.
TX 304
Hillsdale, NJ
*Registered Trademark of E.I. DuPont
8.0 PROCESSING FACILITY AND ESSENTIAL EQUIPMENT REQUIREMENTS
Encapsulation of high voltage power supplies shall require the use of a clean facility to control contamination (espe-
cially silicones, hydrocarbons, vinyl plasticizers and particulate contamination). The clean facility shall have a GSFC
project approved cleanliness level and good ventilation, good lighting temperature control between 20° and 25°C, and
relative humidity ranging from 30% to 60%.
Work space areas must be sufficiently large to contain the actual encapsulating work, essential equipment, instruments,
and all other required hardware. These items shall be inside the clean facility to minimize contamination during the
transfer of flight hardware from one work area to another.
Essential equipment shall include:
a. A calibrated, laboratory weighing balance, accurate to 0.01 gram, used to control material mixing propor-
tions.
b. A vacuum chamber or bell jar capable of operating at a vacuum level 50 microns of mercury (Hg) or less
when empty, used to ensure a bubble free end item.
c. A laminar flow bench clean work station providing a 100 cleanliness class level (per FED-STD-209B) used
to prevent contamination. This work station includes a minimum lighting capability of 100 foot candles of
illumination (1,076 lumens per square meter).
d. A health and safety approved exhaust fume hood for exhausting chemical fumes and vapors used to meet
OSF1A requirements. A minimum of 125 ft/min (38 m/min) air exhaust at the hood face is recommended.
e A clean, oil free, 0 to 100 psi (0 to 0.69 MPa), nitrogen gas supply system including hoses, valves, a regulator
and nozzles used to spray clean and blow dry cleaned surfaces. No flexible polyvinyl chloride (PVC) hose
such as Tygon shall be allowed. Clean nylon, Teflon, polyethylene and Viton hoses are recommended and
acceptable.
f. A lighted magnifier with lOx magnification used for visual inspections.
g. Shore A and D hardness testers, used for verifying proper cured hardness of the encapsulating resin and
epoxy coating.
h. A clean, forced-air convection oven, used in the final process to dry surfaces and other non-flight hard-
ware.
i. A customized vacuum chamber capable of operating at 10 2 Torr or better, when empty. The vacuum chamber
will be used for encapsulating the electronic assembly under vacuum.
j. A spray apparatus for cleaning surfaces and other hardware.
k. A micrometer for obtaining thickness measurements of the reference primer coating.
l. A wet film thickness gauge used to determine the primer application thickness.
Chapter 3: High Voltage Packaging for Space
3-21
High Voltage Power Supply Design Guide for Space
The following list details the model or part number, size or capacity, and source addresses for the essential equipment:
(This representative list is not meant to constitute an endorsement of these products, only that they were used in the
development of this work for space flight hardware on GSFC programs.)
a-1 Mettler Balance
PC4400
Mettler Instrument Corp.
Hightstown, NJ 08520
b-1 Hotpack Thermal Vacuum Chamber including an
18 CFM Mechanical Pump
Floor Model 273700-13
Hotpack Corp.
Philadelphia, PA 19154
c-1 Class 100 Laminar Flow Bench
Model 530-CS
Laminaire Corp.
Rahway, NJ 07065
d-1 Slimline External Air Induction Exhaust Fume Hood
Model ESS-06
Duralab Equipment Co.
Brooklyn, NY 11236
e-1 Nitrogen Gas (99.95% pure or boil off to meet this specification)
Pressure Range: 0 to 100 psi (0 to 0.69 MPa)
Local Sources
f-1 Luxo Illuminated Hand Magnifier, 10X
Catalog No. L-03882-40
Cole-Parmer Instrument Co.
Chicago, IL 60648
g-1 Hardness Testers, Shore A & D
Shore Instruments & Mfg. Co.
Freeport, NY 11520
h-1 Blue M Convection Oven
Model POM-588C-3X
Blue M Electric Co.
Blue Island, IL 60406
i-1 Key High Vacuum Products, Inc.
Customized Chamber
Nesconset, NY 11767
3-22 Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
j-1 Binks Model 15 Spray Gun
Binks Mfg. Co.
Franklin Park, 1L 60131
k-1 Micrometer, 1 inch (25.4 mm)
Catalog No. 12-126
Fisher Scientific Co.
Pittsburgh, PA 15219
1-1 Wet Film Thickness Gauge
Model 115, Part No., GR 6601
Pacific Scientific Co.
Silver Spring, MD 20910
9.0 POWER SUPPLY CLEANING PROCESS REQUIREMENTS
Starting at this point in the procedure, all of the following processes shall be performed in a clean facility. Personnel
shall wear clean facility approved clothing, clean facility approved gloves and follow strict clean facility approved
procedures. Only clean facility approved materials and wiping accessories (see Section 13.0) shall be allowed in the
clean facility.
Thorough cleaning of the high voltage power supply is required to ensure that there is maximum adhesion between
the insulative material and its contact surfaces. In addition to cleaning the power supply, brushes, spatulas, beakers,
cleaning spray apparatus, and drip containers are cleaned at this time to minimize any possible contamination from
these non-flight items.
It is important that power supplies containing electrostatic sensitive parts are protected against ESD. Protection
procedures must be followed during the complete process. All personnel and work stations shall be well-grounded to
prevent any damage during the entire processing of electronics. Wrist straps connected to an approved ESD ground
shall be used. The wrist straps must be worn with the metal contact against the wearer’s skin. Wrist strap electrical
continuity shall be verified daily prior to the encapsulation work. Wrist straps shall be approved by the project or the
Quality Assurance section of the organization performing the work.
CAUTION
Office of Safety and Flealth Administration (OSF1A) precautions and guidelines shall be followed when handling
chemicals and/or organic materials.
Personnel may need to wear a face mask with an organic filter throughout the cleaning and encapsulation process for
protection from solvent and organic resin vapors. Personnel must wear a face mask if required by their local health
and safety standards. The cleaning shall take place under the exhaust fume hood.
Personnel performing the encapsulation work should verify the maximum operating voltage of the power supply and
place that value on the certification log accompanying the power supply.
The substrate surface preparation shall be as follows:
a. Brush and spray clean the following items: primer witness sample substrate, spatulas, blades, containers, drip
container, and any other non-flight items being used in the processing of flight hardware. Use a 1:1 volume
mixture of alcohol/heptane and a precleaned sable brush. Brush clean each item for 2 to 3 minutes then spray
clean each item for 1 to 2 minutes using a Binks Model 15 spray apparatus. Set the spray apparatus between
Chapter 3: FIigh Voltage Packaging for Space
3-23
High Voltage Power Supply Design Guide for Space
40 and 50 psi (0.27 and 0.34 MPa). This ensures that the flow of solvent forces loose any particles adhering
to the items and flushes away contaminants. Use a drip container to catch excess solvent. Dry the items per
steps e and g shown below. Store the cleaned items, unless immediately needed, in a clean bag or cabinet
free of contaminants such as silicones, vinyls, hydrocarbons and particulates.
b. Clean the high voltage power supply. Begin by gently brushing all of the critical surfaces and electronic
parts, including the interior and exterior surfaces of the power supply container. Use a soft sable brush 3/8
inch (9.5 mm) diameter trimmed to 1/4 inch (6.3 mm) in length. Brush the surfaces and parts using a 1:1
part by volume of alcohol/hexane unless directed otherwise. This will loosen any contamination from the
surface and be flushed away when spray cleaned. Use a small soft conductive nylon Gordon brush when
working with an ESD sensitive electronic assembly. The brush should be grounded when used to clean the
assembly.
c. Spray clean the power supply with a Binks Model 15 spray apparatus to clean all surfaces. During the spray
cleaning, hold or position the power supply so that the long length is horizontal and parallel to the cleaning
table top with the open cavity facing the operator. Tilt the power supply 10° to 20° from the vertical so that
the top side is closest to the operator. This will direct the solvent spray away from the operator and towards
the drip container. Spray the power supply back and forth from top to bottom to allow any contamination to
flow downward and off the assembly.
d. Rotate the assembly a full 360° in steps of 90°. On each 90° rotation, spray clean as in step c. Spray clean
each direction for 2 to 3 minutes for small power supplies and 3 to 5 minutes for large power supplies, being
sure to thoroughly spray clean all surfaces and components. Set the spray apparatus gas system pressure
between 40 and 50 psi (0.27 to 0.34 MPa). This will ensure that all surfaces and parts have been thoroughly
spray cleaned. The cleaning of the assembly shall also include exterior wall surfaces.
e. Remove excess solvent from the assembly surfaces by blow drying the cleaned surfaces with dry, clean
nitrogen. Set the nitrogen gas pressure between 40 and 50 psi (0.27 to 0.34 MPa). Dry the power supply in
each direction for 2 to 3 minutes (rotating as in step d), until all of the solvent evaporates.
f. Inspect the assembly surfaces and parts for cleanliness. Use a hand held 10X magnifier. If any stubborn con-
tamination deposits are noted, they should be removed and the power supply recleaned as in steps b through
e. Use precleaned urethane swabs lightly dampened with acetone to remove any stubborn contamination.
g. Dry the cleaned assembly. This final drying removes all cleaning solvent and moisture. Place the assembly
in a clean convection oven at 65°C and dry for a minimum of 30 to 45 minutes.
h. Remove the assembly from the oven and allow it to cool to ambient temperature on the laminar flow bench.
Store the assembly, unless immediately needed, in a clean plastic bag. Place the assembly in a static control
bag if it is ESD sensitive.
i. The assembly should be redried if stored for longer than 18 hours. Repeat step g above prior to the encapsulant
material application.
j. The assembly is now clean and ready for the next step. It is very important that all the surfaces in the power supply
cavity that are to be encapsulated be extremely clean and dry.
10.0 PRIMING REQUIREMENTS
To enhance the adhesion of the encapsulating resin to surfaces and parts, it is critical that primer be applied.
A person who is familiar with the electrical design, the encapsulant resin, and primer should assist the power supply
designer to specify surfaces and parts to be primed. No Shore hardness test sample is required of the primer material.
A primer witness sample using an aluminum sheet, (specified in Section 7.0), should be used to verify proper cure
and allow personnel to continue the encapsulating process with a high degree of confidence that the primer is cured.
The processing shall not continue if the primer exhibits improper curing.
3-24
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
The primer application shall take place at the laminar flow bench or the air exhaust fume hood to control contamina-
tion. The primer shall be kept free of any contamination because contamination may reduce adhesion or dielectric
insulation properties of the encapsulated end items.
Personnel may need to wear a face mask with an organic filter during the priming process at the laminar flow bench
or fume hood because of the primer vapors.
The surfaces and parts, including the primer witness samples, should be coated using a clean, small sable brush.
The primer process shall be as follows:
a. Mask-off any holes on the outside of the power supply assembly container, through which the encapsulant resin
may leak or flow. Use 1/2 or 1 inch (12.7 or 25.4 mm) wide Teflon tape. Criss-cross the tape to ensure that it will
not separate or delaminate during the encapsulation process. Press down firmly on the tape to ensure good adhe-
sion.
NOTE: Each type of encapsulation resin and surface requires its own primer. For the polyurethane, Conap EN-11,
PR 420 works best, except on glass coated surfaces.
NOTE: High Voltage power supply parts with glass coated surfaces must be primed with an Epon/Versamid mix-
ture at least 24 hours prior to the PR-420 primer coating. The epoxy primer application is being described after
the PR -420 because ordinarily few or even no glass coated parts are used in high voltage power supplies.
b. The primer formulation, all materials ±0.05 grams, shall be as follows:
Material Function PBW *
PR-420-A Resin 3.0 g
PR-420-B Curing Agent 21.9 g
• Parts by Weight
NOTE: Part B must be well stirred for 3-5 minutes prior to using.
NOTE: Throughout the entire process, use only plastic spatulas for stirring in plastic beakers. A steel spatula may
scrape plastic off the interior of the beaker and cause contamination. Plastic or stainless steel spatulas may be used
for stirring in glass beakers.
c. Weigh out the PR-420-B under the air exhaust fume hood. Use a laboratory balance. Measure the amounts
specified above. Place the material in a clean glass or plastic beaker.
d. Weigh out the PR-420-A under the air exhaust fume hood. Using the laboratory balance, measure the amount
specified above. Add this to the PR-420-B already in the beaker. Stir well for 3 to 5 minutes.
e. Allow the mixed PR-420 to stand at ambient temperature for a minimum of 15 minutes before using. This
allows entrapped air bubbles to escape. Place a tight lid or cover on the container to prevent evaporation.
NOTE: Do not vacuum de aerate the above mixed PR-420 primer.
f. The following surfaces and parts of the power supply shall be primed as follows:
• High voltage container cavity side walls and bottom — Do not prime unless required.
• Caddock High Voltage Resistor — Prime the resistor body, including the electrical lead wires. NOTE:
The Caddock resistor shall not contain a silicone treated surface when purchased.
• Maida High Voltage Ceramic Capacitor — Prime only the electrical lead wires, including approximately
1/8 inch (0.63 mm) of the capacitor legs. NOTE: Capacitor body shall not contain a wax-treated surface
when purchased.
• Bare electrical wires — Prime
Chapter 3: High Voltage Packaging for Space
3-25
High Voltage Power Supply Design Guide for Space
• Carbon resistors — Prime
• Reynolds connector — Prime
• Electrical insulated lead wires — Prime
• Electrical feed-through terminals — Prime except for glass coated feed-through terminals. An epoxy
coating is used for glass surface feed-through terminals — see step k below.
• High Voltage Cable — Prime if within high voltage area.
• Diodes — Prime except for glass body diodes. An epoxy coating is used for glass body diodes — see
step k.
• Solder Ball joints — Prime
• Aluminum Witness Sample — coat for thickness and cure determination.
• Parts and surfaces requiring primer as specified by engineering drawing and not listed here, shall be
primed.
NOTE: The above recommendations represent the majority of parts and surfaces found in power supplies. Power
supplies vary in design; complexity and parts other than listed above may be found in power supplies. In most cases,
these should be primed if they are within the high voltage power section.
g. Prime the power supply parts and surfaces using a 1/8 or 1/4 inch (3.1 or 6.3 mm) clean sable brush. Apply a
uniform thin coating of the primer to the parts and surfaces, and to the aluminum witness sample. Leave a small
part of one corner of the witness sample uncoated for aluminum thickness determination. Apply the primer to the
parts, surfaces and witness sample to obtain a final dry thickness of 0.001 to 0.002 inches (0.025 to 0.050 mm).
Use a wet thickness gauge to measure the thickness of the coating. Allow the power supply to stand at ambient
temperature between 20° and 25° for 60 minutes before performing the primer acceptance criteria.
h. The primed assembly shall meet the following acceptance criteria when viewed with a lighted magnifier
having 10X magnification, before the encapsulation process begins:
1) There shall not be any entrapped air bubbles in the applied primer surface.
2) There shall not be any unprimed spots on the assembly as required in step g and/or the engineering
drawings.
3) There shall not be any excess primer due to drops, spills or other mistakes on surfaces of the assembly
which are not part of the primed area.
4) There shall not be any primer bridging on any of the parts.
5) The primer shall exhibit proper cure; this shall be verified using the primer witness sample.
6) The primer shall exhibit proper thickness; this shall be verified using the primer witness sample.
Failure of any of the above six acceptance criteria shall be cause for rejection and removing the primer coat. The
primer coat shall be removed and recoated per steps a to g. The primer witness test sample acceptance criteria for
proper primer cure shall be as follows: after 60 minutes, the primer shall be dry and tack free to the touch of a plastic
probe.
The primer witness test sample for proper thickness shall be measured as follows: after 60 minutes, the dry primer
shall not exceed 0.002 inches (0.050 mm) as measured with a micrometer. The measured thickness shall be recorded
in the appropriate documentation attached to the flight hardware.
i. The assembly is primed and ready for encapsulation. The encapsulation shall be performed any time after
the 60 minute stand at ambient temperature between 20° and 25°C, but must be done within a 5 hour time
frame from the time of primer application.
j. Label the primer witness test sample with the flight assembly serial number, date, primer material type and lot
number to ensure material traceability. Store the witness sample at ambient temperature between 20° and 25°C in
a clean bag. The primer witness sample should be labeled and retained for at least one year after the space flight
launch.
3-26
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
NOTE: High voltage power supply parts with glass coated surfaces must be primed with an Epon/Versa-
mid mixture at least 24 hours prior to the PR-420 primer coating. The epoxy primer application is being
described after the PR-420 because ordinarily few or even no glass coated parts are used in high voltage
power supplies.
k. Glass coated feed-through terminals and glass coated diodes shall be primed with a thin Epon/Versamid coating.
Power supplies may or may not have these parts. Any high voltage power supply parts which does not have glass coated
surfaces shall proceed from here to Section 11.0.
The epoxy coating formulation, all materials ± 0.05 grams, shall be as follows:
Material Function PBW*
Epon 815 Resin 10.0 g
Versamid 140 Curing Agent 10.0 g
*Parts By Weight
l. Weigh out the Epon 815 under the air exhaust fume hood. Use the laboratory balance. Measure the
specified above. Place the material in a clean beaker.
m. Weigh out the Versamid 140 under the exhaust fume hood. Use the laboratory balance. Measure the
specified above. Add this to the Epon 815 already in the beaker. Stir well for 3 to 5 minutes.
n. Place the beaker containing the mixed epoxy in a vacuum chamber. Vacuum deaerate at an ambient tempera-
ture between 20° and 25°C at a pressure of 100 microns or less for 10 to 15 minutes or until all entrapped
air bubbles are removed.
o. Remove the beaker and its contents from the vacuum chamber after clean air or nitrogen back fill. Coat the
electrical glass feed-through terminals, glass diodes and the aluminum witness sample using a 1/8 inch (3.1
mm), clean sable brush. Brush the surfaces gently until a very thin epoxy coating has been obtained. The
final smooth and uniform wet coating thickness shall be controlled between 0.0005 to 0.00 1 inches (0.012
to 0.025 mm) thick. Leave a small portion of one corner of the aluminum substrate for thickness measure-
ment verification. The operator should practice resin thickness control prior to the actual resin application
to obtain the specified thickness. Use a thickness gauge to measure the wet thickness of the epoxy coating.
Measure the thickness of the resin on the witness sample.
p. Place the power supply, the hardness test sample, and The witness sample in the vacuum chamber and vacuum
deaerate at an ambient temperature between 20° and 25°C at a pressure of 100 microns of Hg or less for 10
minutes to remove entrapped air bubbles.
q. Excess resin from each mixed batch shall be retained as a hardness test sample; the test sample shall be
processed with the power supply and then used to verify the Shore D hardness. The hardness test sample
shall be 0.250 inch (6.3 mm) thick and approximately 1.5 inches (38 mm) in diameter. The Epon/Versamid
mixture has a pot life of 60 minutes. A fresh mixture should be prepared after this time.
r. Remove the power supply, the witness sample, and the hardness test sample from the vacuum system after
proper clean air or nitrogen gas chamber back-fill. Place the items on the laminar flow bench top surface and
allow to stand at ambient temperature between 20° and 25°C for 18 hours.
s. Place all the items mentioned in step r in a convection oven. Bake at 65°C for 6 hours.
t. Remove the items from the convection oven and place them on the laminar flow bench top. Allow them to
cool to ambient temperature between 20° and 25°C for 60 minutes.
u. Measure the hardness test sample with the Shore Durometer. The hardness shall be a minimum of D-50. Proceed
with the assembly encapsulation if the Shore hardness value is met. The encapsulation shall not be done if the
Shore hardness value is not met.
amount
amount
Chapter 3: High Voltage Packaging for Space
3-27
High Voltage Power Supply Design Guide for Space
NOTE: The hardness test value of the epoxy primer must be determined prior to the final cure on the
flight hardware. After the final cure, as described in step s, it would be difficult to remove without dam-
age to the hardware.
11.0 ENCAPSULATING RESIN APPLICATION REQUIREMENTS
In the encapsulation process, the primed assembly electronics are encapsulated in a vacuum chamber.
The container for the electronic assembly encapsulation is usually no more than 2 to 3 inches (50 to 75 mm) deep.
The container should not present sharp comers or complex surfaces to the encapsulant. In no way should there be any
attempt to restrict the encapsulating material on all sides as it will not yield during thermal excursions, if no room for
expansion has been allowed. See Figure 1 [3.A.I.1] for an illustration of a typical high voltage power supply without
the encapsulating resin.
The chamber used to introduce the polyurethane resin under vacuum should have the following features: A mechanical
pump with a zeolite trap between the mechanical pump and the vacuum chamber to prevent hydrocarbon vapor from
back streaming into the chamber. The chamber should be capable of obtaining a pressure of 10 3 Torr, and equipped
with a vacuum gauge capable of reading the vacuum pressure, a variable back fill gas leak valve, a rotating table in
the interior and workable from the exterior, a vertical stand with a ring holder for the resin cup in the interior and
workable from an exterior control to tip the cup. See Figure 2 [3.A.I.2] for an illustration of the customized vacuum
chamber for encapsulating power supplies.
Excess resin material from each mixed batch shall be retained as a hardness test sample; the test sample shall be used
to verify proper material cure and Shore hardness. The hardness test sample shall be a minimum of 0.250 inch (6.4
mm) thick and approximately 1.5 inches (38.1 mm) in diameter.
It is important that the mixing of the resin formulation and process be performed expeditiously to prevent a significant
increase in viscosity which in turn makes the encapsulating resin more difficult to pour.
The resin shall be prepared in the fume hood and kept free of any contamination because contamination may reduce
adhesion or dielectric insulation properties of the encapsulated end item.
Personnel may need to wear a face mask with an organic filter during the resin preparation at the exhaust fume hood
because of the resin vapors. A face mask shall be worn if required by their local health and safety standards.
The encapsulating process shall be as follows:
a. To prevent resin from overflowing or splattering from the power supply during the process attach a 1 or 2 inch
(25.4 or 50.8 mm) wide strip of Teflon tape around the top side of the power supply’s open cavity. Overlap the tape
approximately 0.250 inch (6.4 mm) on the container exterior wall surface. Press down firmly on the tape surface
attached to the container to prevent lifting. Mask-off any other surfaces or holes to prevent resin contact during
the process. Place the power supply on the flat rotating table in the vacuum chamber in a horizontal position. See
Figure 3 [3.A.I.3] for the power supply mounted on the rotating table in the vacuum chamber.
b. The encapsulating resin formulation, all materials ± 0.05 gms, shall be as follows [EN-1 1-A is no longer available;
use EN-4-A instead].
Material
Function
PBW*
Conathane EN-1 1-A
Resin
100.0 g
Conathane EN-ll-B
Curing Agent
55.0 g
*Parts By Weight
NOTE: Part B curing agent must be well stirred for 3 to 5 minutes before using.
3-28
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
c. Weigh out the EN-ll-A resin under the exhaust fume hood. Use a laboratory balance. Measure the amount speci-
fied above. Place the material in a clean beaker.
d. Weigh out the EN-ll-B curing agent under the exhaust fume hood. Use a laboratory balance. Measure the amount
specified above. Add this material to the EN-ll-A already in the beaker. Stir the material well with a spatula for
3 to 5 minutes.
NOTE: The amount of resin to prepare will depend on the size of the power supply. More than one mix
may be needed to fill large power supply cavities.
e. Place the beaker and the resin mixture in a clean laboratory vacuum chamber. Vacuum deaerate at an ambient
temperature between 20° and 25°C at a pressure of 100 microns of Elg or less for 4 to 6 minutes. This chamber
removes the initial entrapped air bubbles from the resin. The resin should be bubble free at this time.
f. Remove the beaker and its contents from the vacuum chamber after proper clean air or nitrogen gas back-fill.
g. Mount the beaker and the mixed resin in the customized vacuum chamber’s ring clamp on the rotating tippable
stand. Secure well with Teflon tape. Wrap the tape several times around the beaker’s mid- section, e.g., top to bot-
tom, being sure to pass the tape across the ring clamp surface.
h. Position the power supply beneath the resin cup so that it will pour into the middle of the power supply’s cavity.
The power supply may need to be rotated ± 30 degrees during the resin pouring (depends on the size of the power
supply) so that all of the resin is not poured in one place.
i. Place the vacuum chamber bell jar over the resin cup, the power supply, and on the vacuum base plate. Start the
pump and begin the vacuum pump down. Allow the vacuum deaeration to continue for approximately 20 minutes,
before pouring, to extract as much entrapped air and water vapor from the electronics as possible. It is desirable
for the vacuum chamber pressure to reach 1 Torr or less in 5 to 10 minutes, depending on the size of the power
supply and the bell jar size.
j. Pour the resin slowly into the power supply cavity by tipping the beaker with the external control knob. Allow the
resin to slowly enter the power supply cavity until approximately half full. Rotate the power supply platform using
the external rotating knob to spread the resin across the assembly electronics during the pouring. Stop pouring and
allow the vacuum de aeration to continue for approximately 15 to 20 minutes. There will be much resin bubbling
as entrapped air escapes from surfaces and parts. The bubbling becomes less active and slows with time, but may
never completely quit. See Figure 4 [3.A.I.4] for illustration of resin being poured into the power supply cavity in
the vacuum bell jar.
k. After the above 15 to 20 minutes, slowly pour the resin again by tipping the resin cup with the external rotating
knob, until the power supply is almost full and the electronics are covered, or according to engineering drawing
requirements. Continue to vacuum deaerate for another 15 to 20 minutes.
l. Stop the vacuum pumping by closing the mechanical pump foreline valve. Allow the power supply to stand at
vacuum pressure for 5 to 10 minutes for any remaining air bubbles to escape. Use the variable gas valve to slowly
back-fill the vacuum chamber with clean air or nitrogen gas.
m. Remove any accidental particle contaminant or trapped air bubbles in the resin at this time. This shall be performed
within 75 minutes after the resin is mixed. Use a clean needle to pop any trapped air bubbles or to remove any
particles of contamination. Use a lighted 10X magnifier to inspect and find the anomalies.
n. This completes the encapsulation of the power supply. Cure the resin at atmospheric pressure. See Section 12.0
for resin curing.
Chapter 3: High Voltage Packaging for Space
3-29
High Voltage Power Supply Design Guide for Space
12.0 ENCAPSULATING RESIN CURING REQUIREMENTS
The resin should be considered in a partially cured state 48 hours after the encapsulation work has been completed
when cured at an ambient temperature between 20°C and 25°C.
The resin should be considered in a fully cured state 10 days after the encapsulation work has been completed when
cured at an ambient temperature between 20°C and 25°C, or 5 days when cured at a maximum of 40°C but only after the
completed power supply has been allowed to stand for 48 hours at ambient temperature of between 20°C to 25°C.
13.0 ACCESSORIES CLEANING REQUIREMENTS
Cotton swabs, urethane foam swabs and cotton wipes are the wiping accessories used in the clean facility during the
application process. Most of these items are used for cleaning, but all of these items are potential sources of clean
facility contamination if they are not certified for clean facility use.
To be considered certified and suitable for clean facility use, the items listed above require cleaning by a Soxhlet
extraction method for 48 to 72 hours. Cleaning the items is time consuming, however, it is essential that these strict
cleanliness procedures be followed when working with space flight hardware.
Prior to cleaning any material which is to be used in a clean facility, a solvent filled Soxhlet extractor system shall
undergo a thorough, internal, self-cleaning for about 18 hours of cycling. The used solvent shall be properly discarded
upon completion of the internal cleaning, and the system shall be filled with fresh, clean solvent in preparation for the
extraction of the materials which are to be used. Fresh, clean solvent shall be totally substituted for the used solvent
aft every third 48 to 72 hour material extraction period in the process. Glass beads should be included in the flask for
smooth boiling action.
See Figure 6 [3.A.I.6] for a diagram showing two extractors (a 2 or 3 liter glass system recommended) used simulta-
neously to decrease cleaning times. A single Soxhlet extractor may be used in place of the two extractors shown. The
extractors are set up in the clean facility.
The cleaning solvent used in Soxhlet extraction is ethyl or isopropyl alcohol. Once extracted with the alcohol, the
cleaned items are spread out under a clean operating exhaust fume hood for several hours to air dry. A final drying in
a convection oven at 50° to 60°C for 2 hours will complete the process. The cleaned and dry items are stored, unless
used immediately, in a bag or cabinet free of contaminants until needed.
To be certified for clean facility use, the extracted items shall have been cleaned in the Soxhlet extractor system for
48 to 72 hours. In addition, as part of the certification process, three of the extracted items are chosen as test samples
and their cleanliness is verified. Cleanliness of the items is verified by immersing and thoroughly rinsing the cleaned
items in a glass dish containing an alcohol bath, removing the items to allow the alcohol to evaporate, and then test-
ing the alcohol bath residue to determine cleanliness. Testing of the extracted residue is performed through infrared
analysis. To be certified for clean facility use, an Infrared Spectrum of the residue extracted from 3 randomly selected
clean samples shall exhibit a transmission curve of 95% minimum throughout the 2.5 to 16 micron range.
An alternative to using the Soxhlet extractor system is to purchase clean facility approved materials. Approved ma-
terials can be purchased from the Paramax Corporation, Lanham, MD, or from the Coventry Manufacturing Co.,
Baldwin Park, CA. Only swabs and wipes from their Diamond series are acceptable. A certificate of cleanliness level
should accompany the cleaning materials.
NOTE: The above test, relative criteria and acceptance limits on nonvolatile residue from wipes and swabs is a method
used by the Materials Branch. It does not conform to any standard methodology. Alternative methods may be used if
approved by the Materials Assurance Office, Code 313.A.
3-30
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
14.0 APPLICABLE DOCUMENTS
OSHA Standards 29 CFR, Part 1910: Occupational Safety and Health Administration
ASTM D 2240: Standard Method of Test for Rubber Property Durometers Hardness
MIL-STD-1209B: Clean Room and Work Station Requirements, Controlled Environment
15.0 ABBREVIATIONS
CofC
Certificate of Compliance
CVCM
Collected Volatile Condensable Material
ESD
Electrostatic Discharge
GSFC
Goddard Space Flight Center
MSDS
Material Safety Data Sheet
PBW
Parts by Weight
PVC
Polyvinyl Chloride
SMC
Standard Materials Certification
TML
Total Mass Loss
VPM
Volts Per Mil
16.0 FIGURES AND TABLES
The figures on pages 32 through 36 are discussed in this document. Refer to page 32 to review Table 1 [3.A.I.1],
[These original page numbers refer to Figs. 1-6 in the original document and Figs. 3.A.1.1-3.A.I.6 here.]
17.0 ATTACHMENTS
The referenced attachments are located at the end of this document.
Chapter 3: High Voltage Packaging for Space
3-31
High Voltage Power Supply Design Guide for Space
Table 1. [3.A.I.1]. Properties reference of the polyurethane encapsulating resin and the epoxy primer.
Material
Outgassing c
%TML %CVCM
Dielectric
Strength d
VPM
1/16 inch (1.5mm)
at 25°C
Glass Transition
Temperature, Tg
EN-ll-A/B 3
0.68 0.01
610
-65°C
EPON
815/V140 b
0.87 0.05
N/A
N/A
Material
Volume Resistivity
Q-cm at 25°C
Dielectric Constant
at 100 Hz
at 1 MHz
at 25°C
EN-ll-A/B a
4.3 x 10 15
3.30
2.90
a. EN-11 cured at ambient room temperature for 10 days.
b. Epoxy/Versamid cured at ambient temperature for 18 hours, plus 6 hours at 65°C.
c. ASTM E-595, (Tested at 125°C, 24 hours, ICE 5 Torr.) Outgassing data control numbers GSC 10592
and GSC 21214
d. Conap Technical Data Sheet, Bulletin P-156
NOTE: The outgassing data for the PR-420 primer is not provided. Primers of this type generally do not pass
outgassing. In this case, the primer is covered with encapsulation resin.
Solder Joint
Figure 1. [3.A.I.1]. An assembled high voltage power supply without resin encapsulation.
3-32
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Vacuum Chamber
Bell Jar Platform
Back Fill
Gas Valve
Mechanical Pump
Exhaust Hose
Molecular Sieve
Movable Cart
Vacuum Bell Jar
High Voltage
Power Supply
External Knob
for Rotating
Resin Container
External Knob
for Rotating
Turn Table
Vacuum Gauge
Mechanical Pump
Figure 2. [3.A.I.2], A customized system used to encapsulate high voltage power supplies in vacuum.
Chapter 3: High Voltage Packaging for Space
3-33
High Voltage Power Supply Design Guide for Space
Figure 3. [3.A.I.3]. A vacuum system with a turn table and rotating resin pouring container.
3-34
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Turn Table
Vacuum Chamber
Bell Jar Platform
Back Fill
Gas Valve
Mechanical Pump
Exhaust Hose
Molecular Sieve
Movable Cart -
— High Voltage
Power Supply
Rotating Resin Container
External Knob
for Rotating
Resin Container
External Knob
for Rotating
Turn Table
Vacuum Gauge
Mechanical Pump
Figure 4. [3.A.I.4], An illustration of the encapsulation resin being poured under vacuum.
Chapter 3: High Voltage Packaging for Space
3-35
High Voltage Power Supply Design Guide for Space
Figure 6. [3.A.I.6]. Soxhlet extractor assembly — two units.
3-36
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Appendix II to Chapter 3
PROCESS SPECIFICATION FOR PARYLENE CONFORMAL
COATING OF THE CASSINI/CAPS
HIGH VOLTAGE POWER SUPPLY
CAPS Process Specification, CAPS-PROC-313-1, Revision
Goddard Space Flight Center
Chapter 3: High Voltage Packaging for Space
3-37
High Voltage Power Supply Design Guide for Space
Parylene Conformal Coating of the Cassini/CAPS
High Voltage Power Supply
1. SCOPE
This document establishes the procedure to be used for conformal coating of the Cassini/CAPS High Voltage Power
Supply (HVPS) with Parylene C.
2. APPLICABLE DOCUMENTS
SPECIFICATION
Goddard Space Flight Center
GC1308729E Box, HVU-1 Power Supply
3. REQUIREMENTS
3.1 Equipment
a. Beaker, glass, 4-liter, from Fisher Scientific or equivalent
b. Polyethylene container, 1-gallon, from local sources
c. Teflon-coated magnetic bar, l-inch, from Fisher Scientific or equivalent
d. Bink No. 15 spray gun from Binks
e. Timer from local sources
f. Micrometer, 1-inch, from Fisher Scientific or equivalent
g. Tweezers, stainless steel, from local sources
h. Razor blade from local sources
i. Exacto blade from local sources
j. Brush, V^’-wide, non-coductive Nylon, from local sources
k. Sharp-pointed Teflon standoffs (4), 1 to 1.5-inch high, from Code 313 or equivalent
l. Balance, 500± 0.01-gram capacity, from Mettler Instrument or equivalent
m. Magnetic mixer plate, catalog no. 11-500-7SH, from Fisher Scientific
n. Vacuum oven with pressure of less than 300 millitorr from Blue M Electric Co. or equivalent
o. Convection oven, Blue M Model POM588C-3X, from Blue M Electric Co.
p. Convection oven, clean facility, Blue M, from Blue M Electric Co.
q. Exhaust fume hood in Room 140, Building 30, Code 313
r. Class 100 Laminar Flow Bench, Model 530-CS, from Laminaire Corp.
s. Polaroid camera, MP-4, from Polaroid
t. Optical microscope from Nikon
u. Horizontal Parylene Coating Machine, Model 1293, Paratronix
3.2 Materials
a. HVPS per GC1308729E without the low voltage board, from Code 734.3
b. Parylene C from Specialty Coating Systems
c. Organosilane A-174 primer from Specialty Coating Systems
d. Deammoniated Liquid Maskant TC-530 from Kester
e. De-ionized water with resistance value of 18.2 MO or better from Code 313 or equivalent
f. Reagent-grade isopropyl alcohol (IPA) from Fisher Scientific or equivalent
3-38
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
g. 200-proof ethanol from GSA or equivalent
h. Reagent-grade heptane from Fisher Scientific or equivalent
i. Temp-R-Tape K102 from Furon, CFIR Division
j. Soxhlet-extracted cotton wipes from Code 313
k. Soxhlet-extracted foam swabs from Code 313
l. Latex gloves CR-100 from Baxter
m. Polyethylene gloves from Fisher Scientific
n. Aluminum foil, stock no. 8135-00-724-0551, from Alcan Foil Products or equivalent
o. Dry nitrogen gas, 99.99% pure, from local sources
p. Aclar 22C bags from Clean Room Products
q. Witness coupons, 1” x 2” x 0.001” aluminum strips (7) and l”x2”xl/16” polyethylene strip (7) , from
Code 313
r. Parylene aluminum boats from Code 313
3.3 Procedure for Parylene Conformal Coating
It should be noted that since the low voltage board is removed from the F1VPS before the F1VPS is delivered to Code
313 for the Parylene coating, ESD precaution is not a requirement for these procedures.
Portions of this procedure shall be performed in the Class 10,000 clean room. The Flight Assurance Manager or Qual-
ity Control engineer shall be notified at least twenty-four (24) hours prior to the start of this procedure.
3.3.1 Parylene Coating Equipment and Material Verification
1. Clean and precoat the Parylene Coating Machine with 25-gram of Parylene prior to coating the F1VPS.
2. Verify that the chiller will obtain a temperature of less than -85°C, otherwise, the IPA needs to be replaced.
NOTE:
1. ACS-grade isopropyl alcohol has a flash point of 72°F. The Parylene coating process shall be conducted in
a well-ventilated area.
2. Prolonged breathing of Parylene vapor fumes should be avoided. An organic vapor mask should be worn if war-
ranted.
3.3.2 HVPS Inspection
1. Inspect the HVPS at 4-10X power of magnification for anomalies, imperfections, contaminants, and broken
wires. Any defects shall be reported to Code 734.3 personnel.
2. Take Polaroid photographs of the HVPS prior to coating.
3. 3. 3 Primer Preparation
The following processing steps shall be conducted under the exhaust fume hood in a class 10,000 clean facility.
1. Place a 4-liter beaker on a magnetic mixer plate.
2. Pour into the beaker 1400cc of de-ionized water, 1400cc of IPA, and 28cc of primer A-174.
3. Place a Teflon-coated magnetic stirring bar in the beaker and cover the beaker with a clean piece of aluminum
foil.
4. Turn on the magnetic mixer plate and mix the primer solution for two (2) hours minimum at room temperature.
Chapter 3: High Voltage Packaging for Space
3-39
High Voltage Power Supply Design Guide for Space
3.3.4 Cleaning of the HVPS
The following processing steps shall be conducted under the exhaust fume hood in a 10,000 clean facility.
All supporting hardware and materials to be used in this procedure shall be cleaned with a 1 to 1 part by volume of
200-proof ethanol to reagent-grade heptane, air dried, then baked in a convection oven for thirty (30) minutes mini-
mum at 60°C to 65°C. This solvent mixture will be referred to as ethanol/heptane solvent.
NOTE: Clean polyethylene or latex gloves shall be worn at all times during processing of the transformer. Only
polyethylene gloves are acceptable for processing with a solvent.
The HVPS shall be cleaned under the exhaust fume hood of the class 10,000 clean facility as follows:
1. Brush gently all surfaces of the HVPS to loosen any contaminant present, using a Nylon brush with the
ethanol/heptane solvent.
2. Spray clean the HVPS for two to three minutes with the ethanol/heptane solvent using the Binks no. 15
spray gun at approximately 25 to 35 psi with clean, dry nitrogen gas. The spray gun should be positioned at
a 45° angle from the horizontal surface of the HVPS. Rotate the HVPS 90° from its original position until
the entire HVPS is spray-cleaned. Repeat this step four (4) times.
3. Blow dry the HVPS surfaces with clean, dry nitrogen gas.
4. Dry the HVPS in air for 5 to 10 minutes minimum.
5. Bake the HVPS between 60°C to 65°C for thirty (30) minutes minimum in a Convection oven.
3.3.5 Priming of the HVPS
The following processing steps shall be conducted under the exhaust fume hood in a class 10,000 clean facility.
1. Place the HVPS in a polyethylene container, 1-gallon size or large enough to hold the HVPS.
2. Pour the primer solution (from Paragraph 3. 3. 3.4) in the container until the HVPS is totally submerged.
3. Let the HVPS soak in the primer solution for forty (40) minutes ±1 minutes.
4. Remove the HVPS from the primer solution and place the HVPS on three (3) or four (4) clean, sharp-pointed
Teflon standoffs. The HVPS components shall be facing down.
5. Allow the primed HVPS to air dry for thirty (30) minutes + 5/-0 minutes.
6. Rinse the primed HVPS with IPA for approximately 2.5 minutes +1/-0 minute to remove primer stains.
7. Blow dry the HVPS surfaces with clean, dry nitrogen gas.
8. Inspect the HVPS surfaces for primer stains. If present, repeat steps 6 through 7 of this section as needed
to remove primer stains.
9. Bake the HVPS between 60°C to 65°C for thirty (30) minutes minimum in a convection oven.
10. Place the primed HVPS in a clean Aclar 22C bag and remove it from the clean room for thermal vacuum
bakeout.
NOTE: The HVPS must be coated within twenty-four (24) hours after the primer application.
3.3.6 Moisture Removal
The following processing steps are acceptable in a class 300,000 clean facility.
1. Bake out the vacuum oven at a pressure of less than 300 millitorr at 125°C ±5°C for twenty-four (24) hours
minimum prior to placing the primed HVPS in the oven.
2. Place the HVPS in the vacuum oven with the temperature set between 65°C to 70°C for eight (8) hours
minimum. Set the oven’s fail safe temperature at 75°C.
3-40
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
3. Close and seal the vacuum door. Close the vent valve. Open the vacuum valve.
4. Remove the HVPS from the vacuum oven after bake-out and let it cool down to room temperature.
5. Place HVPS in the Aclar bag and take it to the clean room. This step will continue as indicated in Paragraph
3.3.8.
3.3.7 Preparation of the Parylene Coating Machine
The following processing steps are acceptable in a class 300,000 clean facility.
1. Turn on the chiller, vaporizer, and furnace heaters for at least one (1) hour before coating.
2. Fabricate three (3) Parylene C boats (2” x 8” x 1”) from clean aluminum foil.
3. Fill each boat with 70 g of Parylene C. The 210 g of Parylene C will provide an approximately 1.5 to 2-mils of
Parylene coating of the HVPS.
3.3.8 Masking of the HVPS
The following processing steps shall be conducted in a class 10,000 clean facility.
Masking shall be performed immediately after the bake-out process. The procedure is as follows:
1. Remove the HVPS from the Aclar bag in Paragraph 3. 3. 6.5.
2. Mask the surfaces of the HVPS per Drawing No. GC1308729E and details (Figures 1-3 [3.A.II.1-3.A.II.3])
instructions from Code 734.3’s Cassini/CAPS personnel, using the Temp-R-Tape K102 and Maskant TC-
530.
3. Verify that the masked areas are in compliance with the instruction prior to the Parylene coating.
NOTE: This inspection must be performed by Code 734.3’s Cassini/CAPS personnel.
4. Place the masked HVPS in a clean Aclar bag and remove it from the clean room for Parylene coating.
3.3.9 Parylene Coating of the HVPS
The following processing steps are acceptable in a class 300,000 clean facility.
1. Remove the HVPS from the Aclar bag in Paragraph 3. 3. 8.4. Place the masked and primed HVPS on three
(3) or four (4) sharp-pointed Teflon standoff. The HVPS shall be on the center shelf and in the center of the
Parylene deposition chamber.
2. Label all 18 witness coupons, measure, and record the thickness of each coupon.
3. Place three (3) aluminum witness coupons and three (3) polyethylene witness coupons in the chamber, one
each on the center shelf and, one directly above it. Place one (1) aluminum witness coupon and one (1) poly-
ethylene witness coupon on the shelf directly below the center shelf.
4. Close the chamber door.
5. Place one (1) Parylene C boat (from Paragraph 3. 3.7.3) in the vaporizer tube and close the endcap.
6. Turn on vacuum for 20-30 minutes and verify that the vacuum has reached 5pm or less on the T c gauge and 5pm
on the ion gauge before turning on the vaporizer.
7. Slide the heater over the vaporizer tube to start the vaporizer. The vaporization will start at approximately 135°C
to 140°C. The vaporizer pressure on the ion gauge shall be between 30-40 pm.
8. Continue until the Parylene is depleted. The Parylene can be observed by sliding out the vaporizer heater.
9. Turn off the vacuum valve, and back fill the chamber.
10. Open the chamber door to remove two (2) witness coupons (1 aluminum and 1 polyethylene) from the shelf
directly above the center shelf, where the HVPS is, and from the center shelf.
Chapter 3: High Voltage Packaging for Space
3-41
High Voltage Power Supply Design Guide for Space
01 SO 0
Figure 1. [3.A.II.1], FIVPS bottom view — Mask mounts.
3-42
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
CT> i-
^ 00 LO
Areas to be masked from parylene coating. The ceramic insulators
and output electrodes are to receive coating. The front faces of the
output terminals are to be masked.
Figure 2 [3.A.II.2], Cassini/CAPS High Voltage Unit-1. Output ends detail.
Figure 3 [3.A.II.3]. HYPS, HV-Side
Chapter 3: High Voltage Packaging for Space
3-43
High Voltage Power Supply Design Guide for Space
1 1 . Measure the thickness of the witness samples using a micrometer. Determine the Parylene coating thickness.
Use this thickness to calculate for the amount of Parylene necessary to put on a total of 1.5 to 2 mils coating
on the HVPS and, thus, the number of times to repeat the coating process (see Paragraph 3.3.9.16).
12. Close the chamber.
13. Remove the endcap on the vaporizer tube.
14. Remove the boat and place another boat containing the 70 grams of Parylene.
15. Close the endcap.
16. Repeat steps 6 through 13 of this section as necessary.
17. Verify that the last witness coupons on the center shelf have an approximately 1.5 to 2-mil of Parylene coat-
ing.
18. Remove the HVPS from the chamber.
19. Turn off the chamber controls for the chiller, vaporizer, and furnace heaters.
20. Place the HVPS in a clean Aclar bag and carry it to the class 10,000 clean facility.
3.3.10 Removal of Masking Tape from the HVPS
The following processing steps shall be conducted in a class 10,000 clean facility.
1. Remove the Temp-R-Tape K102 from the HVPS using the Exacto blade to cut the Parylene at the tape edge to
prevent delamination and feathering. Cutting shall be performed under the microscope at 4-10X magnifica-
tion. Use a tweezer to lift the tape while cutting the Parylene. Trim all areas of feathered Parylene by using
a tweezer and an exacto blade.
2. Clean the unmask areas with either extracted wipes or extracted foam swabs that have been dampened in
ethanol. This process will remove tape residue and the primer.
3. Let the HVPS air dry for 5 to 10 minutes.
4. Place the HVPS in a clean Aclar bag and remove it from the clean room.
3.3.11 Preparation for Delivery
The following processing steps are acceptable in a class 300,000 clean facility.
1. Remove the HVPS from the Aclar bag.
2. Take Polaroid photographs of the Parylene-coated HVPS.
3. Place the HVPS back in the Aclar bag.
4. Place the bagged hardware in the original bag the HVPS was delivered in; the later bag shall have been pre-
cleaned with wipes dampened in ethanol and blow dried with nitrogen gas.
5. Complete a Code 313 Work Request for Polymeric Processing.
6. Record the related processing information on the certification log and enclose one aluminum witness coupon from
the final coat indicated in Paragraph 3.3.9.17, and the photographs indicated in Paragraphs 3.3.2.2 and 3.3.11.2.
4. QUALITY ASSURANCE PROVISION
4.1 Inspection
Quality Control shall visually inspect for air bubbles, delaminations, etc., of the Parylene coating. Quality Control
shall also verify the coating thickness of the witness coupons from the final coat indicated in Paragraph 3.3.9.17.
3-44
Chapter 3: High Voltage Packaging for Space
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
5. NOTES
5.1 All cotton wipes and swabs shall be Soxhlet-extracted and oven-dried prior to use. These processes shall be
performed by Code 313 personnel.
5.2 Material Safety Data Sheets (MSDS) shall be reviewed prior to handling of or processing with the solvents and
materials specified in this document.
Chapter 3: High Voltage Packaging for Space
3-45
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 4: High Voltage Parts
General Comment
Identification of commercial products and equipment to adequately specify or document the high voltage design does
not imply recommendation or endorsement, nor does it imply that the product and equipment identified is necessarily
the best available for one purpose.
I. High Voltage Parts: Passive Parts
This can only be a superficial survey, trying to touch on some of the most important and obvious topics since each
type of part is an entire study in itself.
High voltage parts are capacitors, resistors, transformers, diodes, feedthrus, connectors, standoffs, cables, etc. There
are no reliable high voltage relays or switches for space. For space use, the high voltage parts should be solid, without
any trapped gas — if at all possible. Much of the “corona” testing activity is concentrated at the parts level since it
is much easier to identify location and causes of failure at the parts level than in a completely assembled HV power
supply. See Chapter 5.
Specification to which various parts types should be ordered are either Goddard Space Flight Center Preferred Parts
List (PPL) or military specification. Even these do not eliminate all the difficulties that might arise.
A. HV Resistors
HV resistor bodies must be solid, not hollow. They must be of low inductance construction and low temp coefficients.
Some types that have been flown successfully are:
Manufacturer
Victoreen
Type
MOX 1125-23
MOX 750-23
MOX 400-23
Goddard Spec.
S-311-P-672
Caddock
MG (Precision) S-31 l-P-683
and
Low Inductance
Caddock TG (Precision) S-31 l-P-741
Low Inductance and
Low Temp, coefficient S-311-P-742
RPC
BBMW
IRC
RHV
MIL-PRF-49462B
One of the questions that has arisen is whether primer coats before potting affect the precision resistance of high
resistance devices. Obviously this depends on the resistance, but testing carried out with 200 Megohm Caddocks
of both the MG and TG types showed that thin primer coats of PR420 or of Epon 828/Versamid 140 epoxy, which
are good primers for polyurethanes, do not spoil the ±1% precision. Thick primer coats of the epoxy do spoil the
tolerance. It is also important to ensure that after the manufacturer is finished with his testing program where he
Chapter 4: High Voltage Parts
4-1
High Voltage Power Supply Design Guide for Space
uses silicone oil to immerse the resistors, that all traces of silicone oil have been removed by vapor degreasing before
delivery. Otherwise there will be poor adhesion upon potting. Also, low temperature coefficients of resistance are,
of course, very important.
B. HV Capacitors
Again, for space use, the body of the capacitor must be solid. Therefore, thin film high voltage capacitors can not
be used, either the cylindrical hollow types or even the so-called encapsulated flat types which still have air gaps
between film layers.
So, what can be used are ceramic single disc types, or after much testing, (see later), ceramic multilayer types or
resin-impregnated mica paper types.
Some types, that have flown successfully, are:
Manufacturer
Maida
Calramic
Technology LLC,
formerly KD/Center
Electronics
TDK
Type
Single disc ceramic
Multilayer, Monolithic
Ceramic
Single disc Strontium
Titanate, very thick
Custom Electronics Resin impregnated mica paper
(much larger than the Ceramics).
Useful as output filter
caps above 10 kV.
Specification
S-311-P-15C(D) (Goddard)
MIL-PRF-49467B, since May 2001
Flying in FIRS supply on Flubble Space
Telescope
Johanson Dielectrics
Capacitors should be DC or even AC Partial Discharge tested among other tests. They must be derated to 60% of the
manufacturer’s rated voltage; the multilayer capacitors should be derated to 50% of the manufacturer’s.
Maida single disc BaTi0 3 ceramic capacitors (the X5R or X7R formulation) have been flown successfully up to about
10 kV DC for up to about 2000 pF capacitance. When the voltage and/or the capacitance gets much higher, then there
have been instances of failures. This is because when the discs get too thick, the shear stresses of the piezoelectric
material causes cracking. Or when the manufacturer tries to supply Z5U formulation, the capacitance, or C-charac-
teristics are very variable with respect to temperature and voltages.
The coating on the ceramic capacitors must be fluidized-bed epoxy, compatible with potting. This coating must pre-
serve high resistance characteristics at high temperature extremes (meaning at least to 85°C) and must not crack at
low temperature extremes (at least as low as -35°C). Mil specs demand temperature ranges from -55°C to +125°C,
and the problems which ensue from this requirement have prevented ceramic coated capacitors being available to
Mil specs. Potted ceramic caps for high voltages must be avoided, because the manufacturers do not vacuum pot, and
they include objectionable bubbles in the potting, and they also do not put enough potting between the ceramic chip
and the top of the potting shell.
4-2
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
When the high voltage ceramic multilayer capacitors first appeared on the market, some failures and difficulties
were encountered with them. There were strenuous testing programs and “Evaluations.” An example of this is seen
in Appendix 1 to this chapter (Chapter 4), Ref. [1] (1993). By May 2001, experts in the field added an AC corona test
to MIL-PRF-49467B as seen in Ref. [2]. The authors of this HV Guide very much approve of this, contrary to Ref.
[1], ft must be remembered, however, that the M1L-PRF-49467B is (a) for ground-based use where equipment can be
easily repaired, and that it includes (b) manufacturers’ encapsulation which is sometimes not done carefully enough
for space. The multilayer capacitors should be ordered either bare or coated with fluidized-bed epoxy.
Various configurations of plate design in HV multilayer capacitors are reproduced (greatly enlarged) as Figure 4.1 taken
from Ref. [1], The series parallel configuration is more reliable and reduces the electric field at the tips of the plates.
4 Series x 8 parallel
Figure 4.1. Different plate designs in high voltage ceramic multilayer capacitors, Ref. [1].
The design engineer of the HV power supply will still request the HV capacitors to be “screened” or 100% acceptance
tested. The screening contractor has an established set of tests whose sequence and repetition frequency can be modi-
fied slightly. One version of the tests is shown in Table 4.1. This is a somewhat changed version from Ref. [1]. The
screening contractor may also do destructive physical analysis (DPA) on a small sample from a given lot.
Chapter 4: High Voltage Parts
4-3
High Voltage Power Supply Design Guide for Space
Table 4.1. Screening test methods for HV ceramic multi-layer capacitors, adapted from Ref. [1].
Test
Standard/Spec.
Test Condition
0: Visual
M1L-C-123
3 to 1 Ox Magnification
1 : Capacitance
& Dissipation
Factor
M1L-STD-202
For X7R dielectric: 1 V at 1 kHz
ForNPO dielectric: 1 V at 1 kHz
except for C < 1 000 pf, then 1 MHz
Fail: Cap: outside of tolerance
DF: > 2.5% for X7R, and > 0.2% for NPO
2: Insulation
Resistance
MIL-STD-202
500 V DC applied;
Fail: below limit of 100 GQ at 25°C
and below 10 GQ at 85°C
except where C> 10,000 pf, then
66 GQ and 6.6 GQ respectively
3 : Dielectric
Withstanding
Voltage
M1L-STD-202
-Dielectric
-Body
Ramp voltage to 1 00% rated voltage in 1 min.
between the leads. Dwell 5 s max.
Fail: Current > 1 pA.
Ramp to 500 V DC, applied between metal
leads and metal tape wrapped around body
4a: De-age or de-polarize at 85°C for 12 h with leads shorted.
4b: DC Partial
Discharge
Adapted from
ASTMD
1 868-8 1 &
IEEE STD
454-1973
Stepwise ramps & plateau sequences;
the pulse and total charge sum data from
0 to half-rated voltage, half-rated voltage
to lull rated voltage, at the quiescent
plateau, and from rated voltage to 0. See
Chapter 5.
5 : AC Partial
Discharge
Adapted from
ASTMD
1 868-8 1 &
IEEE STD
454-1973
Ramped 60 Hz AC voltage until 1 00 pC
pulses are detected. Dwell there no longer
than 5 s. No measurement above 70%
of the DC rated voltage (equivalent in rms
volts)
6: Temperature
Cycling
MIL-STD-202
5 cycles from -55°C to +85°C for
Cycling screening.
7: Voltage
Conditioning
MIL-C-49467
Ramp voltage to 1 00% DC rated voltage
(DWELL length decided by Design Eng.)
8: 85°C/85%
Relative
Humidity
MIL-C-123A
Electrical measurements; do not exceed
limits above. Capacitance delta not to
exceed 10%. Low voltage test.
Test 4a and 4b or 4a and 5 can be repeated, either at atmospheric pressure with the capacitors immersed in FC-43
Fluorinert liquid, or in vacuum at 10“ 6 torr. For screening, go only to 100% V rated .
4-4
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
A brief mention of the types of designation of ceramic materials, most of which contain a certain amount of BaTi0 3 ,
follows. These designations are X5R, X7R, Z5U, NPO or else BX, COG, etc., and these have to do with AC, capacitance
change due to temperature change. It also follows that the dielectric constants of the different materials are different,
and that the change of capacitance with applied voltage change is different. In addition, confusion arises as to whether
the Specification is MIL-Spec. or EIA. In general, COG equals NPO, and BX includes X5R and X7R characteristics.
It is best to request literature from the manufacturers themselves, such as, for instance Technical Bulletin No. 792,
“Understanding Chip Capacitors” by Johansen Dielectrics, Ref. [3]. To give an idea of the enormous differences of
dielectric constants of ceramic capacitors, see Table 4.2:
Table 4.2. Approximate dielectric constants of ceramic capacitors.
Designation
NPO
BX
X7R
Z5U
Class I
Class II-mid K
Class II-high K
Dielectric
constant
Less than
150
1000-2000
1000-2000
4000-7000
In general, the NPO ceramic has the least percentage change in capacitance with temperature. But also, partial dis-
charge tests, being very revealing for X7R, X5R, and Z5U capacitors, are not very useful for NPOs other than that
no partial discharges may be tolerated at all.
C. Connectors
A connector that is mated on the ground is necessarily filled with air at atmospheric pressure. Then, depending on
the leak-rate from the shell of the connector, where the cable enters or at the mating interface, the pressure of the
air within it will get down to the “critical corona region” at some indeterminate time in orbit. This will then cause
a metal- to -metal (center-pin to grounded shell) continuous gas discharge within the connector. This will happen no
matter how double -backed or convoluted the charged particle path is from high voltage center-pin to grounded metal
shell. Two approaches to prevent this can be tried:
(1) Vent the connector by drilling holes through the shell.
(2) Interrupt the metal-to -metal gaseous path with a tight fitting insulating solid seal. Then one will get only
occasional small partial discharge pulses instead of a continuous current or an arc.
The interrupting solid seal must, of course, be elastic and maintain its elasticity (no thermal set), despite many thermal
cycles from -55°C to at least 85°C. The above discussion is usually talked about in terms of “creepage path” by the
connector manufacturers instead of the “gas discharge” language of the physicists.
Reynolds Industries of Los Angeles, California, developed such an interrupting seal, Ref [4], in the form of a Silastic
O-ring in the receptacle of a single pin connector. This is in their Series 600 connector, which is in addition clad
internally, as well as the nose of the female plug, with Diallyl Pthalate insulator. This connector has been intensively
tested in vacuum in the corona region at Goddard Space Flight Center, Ref [5], Such tests require opening and closing
the connector while inside the vacuum system at corona region pressures, from the outside of the vacuum system via
a movable vacuum penetration. Then, being sure the connector has the worst possible pressure inside, high voltage
is applied and partial discharges and/or continuous current, if any, are measured. Tests have been successful to 6 kV
DC, and many of the Series 600 have flown. See Figure 4.2.
Chapter 4: High Voltage Parts
4-5
High Voltage Power Supply Design Guide for Space
Plug/Contact
~~ 1 — I
— r
\\\\\\\/
J
Diallyl phthalate
a) Female connector P/N 167-3770 A
Receptacle/Pin
b) Male panel connector P/N 167-3771
Figure 4.2
Reynolds Industries wrote a short history of their high voltage connector development. Reynolds has now brought out
Advanced single-pin high-voltage connectors with molded, multiple interface seals in the receptacle male portion of
the connector. However, none of these have been tested at GSFC as described above, and none of these have flown on
GSFC missions. The manufacturer’s ratings claim to go to 25 kV DC on these Advanced connectors.
D.HV Cable
There are many dielectric insulation types used, e.g., polyolefin, Teflon, Kapton wrap, silicone rubber, some shielded
and some not. For voltages above 10 kV, the cable should have “corona strand shielding.” That is, a semiconducting
black material should fill the space between the metal center strand and the inside of the dielectric insulation and
is also coated onto the outside of the dielectric insulation before the metal shielding braid is applied coaxially. The
jacket outside the shielding braid (this braid must be grounded) should adhere to potting, or if made of Teflon, must
be pre-etched.
It must be kept in mind that Teflon and especially silicone rubber cold-flow. Sharp right-angle and hair-pin bends
must be avoided. Whereas silicone rubber HV cable is nicely flexible for work in the laboratory and on the ground,
it should not be used in flight modules because it is more troublesome as to cold-flow and pin-holes than any other
dielectric insulating material.
4-6
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
At the cut end of the cable, the metal shield and the outside anti-corona coating must be peeled back or removed at
least 1.5 inches from the cut. Follow manufacturer’s directions for cable-to-connector assembly, but no shrink tubing
may be used.
Within a HV power supply thick bare bus wire can be used, properly spaced with respect to openings in shielding
partitions and away from grounded terminals. The potting will do the insulating. HV cable is then only needed at
the HV output and proper attention then needs to be given as to how it exits through the shielding box via a HV
feedthrough.
A few additional rules for HV assembly with cables or bus wires are:
1. Solder balls must be used at all HV junctions, and sharp points are prohibited.
2. Even dielectrics and insulating materials should have smooth rounded edges.
3. Corona balls or corona doughnuts are advisable.
4. Parts and wiring should be mounted up off the circuit board with mechanical stress relief wires so that potting
or alternatively coating can get under the high voltage part and not lock in a layer of air under the part. Also
no shrink tubing or sleeving on cables or bus wires are allowed.
5. However, if the entire construction is board-mounted or hybrid construction, such as for voltages below 3 kV, then
parts have to be used and ordered especially for that construction with the suitable terminations. As a specific
example, if ceramic multilayer chip capacitors are to be mounted with the bodies parallel to the board, then only
a very thin coating with preferably Parylene is allowed, leaving a gap of one or two mils between the coated body
of the capacitor and the coated circuit board. Rigid gluing of the body of the capacitor to the circuit board is not
allowed because that would cause it to shear and crack due to thermal stresses when thermal cycling occurs.
E. High Voltage Circuit Boards and HV Standoffs
Most of the parts and components of high voltage power supplies for space applications are mounted on circuit boards.
As already noted in earlier HV guide books, Ref [6], “the most popular board materials are epoxy-impregnated fi-
berglass, types G-10 and G-ll.” Polyimide impregnated fiberglass boards, G-30, are used for higher voltages. When
circuit trails are printed on the G-10 or G-ll boards, they have relatively sharp edges, and also water vapor absorbs
on the surface of the boards. This causes the flash-over voltage along the surface of the board from one trail to the
neighboring one to be about 2 kV/mm at 50% relative humidity and atmospheric pressure. The practical limit for un-
coated boards with reasonable (1 cm) conductor spacing seems to be about 20 kV. This can be improved considerably
by vacuum bake-out, followed immediately by coating with polyurethane or Parylene vacuum deposition.
For the higher voltages, e.g., 25 kV, G-30 boards are used and the parts and connecting wires are mounted up off the
G-30 boards on high voltage stand-offs. The connecting wires on any one board are usually bare bus wire; insulated
cable with shielding is only used to connect one board to another, or as output cable to the instrument. Later potting
of the entire HV portion serves to insulate the bare bus wire. The stand-offs are usually made with porcelain or Teflon
insulators between the top metal solder terminal and the bottom metal screw for attaching to the circuit board. Above
approximately 10 kV, it is desirable to have the bottom screw and the locking nuts made of plastic, such as polycar-
bonate, since the high voltage electric fields will induce high voltage onto them, even though they are insulated from
the top solder terminal. For the same reason, the holes into which the bottom screws are attached, must be vented,
so they can thoroughly outgas. Teflon stand-offs have to be pre-etched for subsequent adhesion to potting or coating.
Every part, such as resistors and capacitors and diodes, etc., has to have its own mechanical stress relief bend in the
connecting leads.
When hybrid type of construction and assembly was first attempted for the lower HV supplies, up to about 2 or 3 kV,
with the bodies of the parts rigidly adhered to ceramic boards with conformal coating, this was at first unsuccessful.
Chapter 4: High Voltage Parts
4-7
High Voltage Power Supply Design Guide for Space
The reason was that special precautions were not taken to prevent thermal stresses during temperature cycling, causing
cracking of solder joints or ceramic boards or even breaking of ceramic multilayer capacitors.
More recently, for voltages below about 3 kV, a so-called Flex Board is being used, made of a layer of 0.002 inches
of Kapton Flex between two layers of rigid polyimide. See the multicolored figure in Chapter 5, Appendix III. It is
undergoing extensive testing. Metallic trails are printed onto the Flex Board; see Figures 7.1.1 and 7.1.2 in Chapter
7, HVBS (I), ADC/GLAST. After assembly, it is Parylene-coated with a 1 mil thick vapor deposition. Great care is
being taken to have several mils spacing between the bodies of board-mounted parts, such as ceramic chips capaci-
tors and diodes, and the Flex Board. In fact, parts such as high voltage diodes, can now be purchased either for axial
mounting with the usual axial wire leads, or for board-mounting with “surface mount square tabs” for soldering to
the trails. The “square tab” serves as a stress relief and prevents the body of the diode from lying directly on the board
surface. Inquiry about the Flex Board and trails was made to Dupont Flexible Circuit Materials in Research Triangle
Park, North Carolina.
II. Transformers
A. Introduction
Faraday’s and Ampere’s laws govern high voltage transformer operations just as in low voltage ones. In addition,
parasitics and distributed parameters play a dominant role because of the high turns ratio and the very large number
of secondary turns. The equivalent circuit is shown in the Figure 4.3 together with some “ball park” values for the
distributed parameters.
The C’s come from the interwinding capacitance. It is also determined by the dielectric constant in which the wind-
ings are immersed, that is K e = 1 if not potted, and K c is somewhere between 2 and 4 depending on the potting com-
pound.
The coupling coefficient between primary and secondary is 90% or better when the windings are wound concentrically
such as in a Ferrite pot core transformer. But the coupling coefficient may only be 60% when gapped Ferrite C-cores or
U-cores are used with windings separated on opposite legs of the core. The leakage inductance thus created gives rise
to L lp and L ls , in Figure 4.3. A converter topology must be chosen to accommodate the large leakage inductances.
The formula for the resonant frequency co 0 (where c) n =2nf,), as shown in the Figure 4.3b of the primary tank network,
is
(0 = and Q=«) n N 2 C,R.
(L M N C S ) 2
This is an example of parallel resonance, and the impedance is highest at resonance; thus, the current drawn is lowest
when driven at resonance. On the other hand, the higher the frequency of operation away from resonance, the lower
the impedance and the input current drawn is higher. This can happen just due to temperature cycling between -35°C
and +85°C which causes the distributed parameters to change.
In general, secondaries above 2 kV, measured from zero-to-peak (0-pk), require special attention. (Some have flown
as high as 10 kV, 0-pk, on UIT.)
4-8
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
L, p ~ 10 (iH Ideal (1:N), N~50
O
L m -100 |iH
o
L ls ~10%x 2500 x 100 |iH
_tyyy^
0.05/|iF
/N 2
(a) First order approximation
o
N 2 C S = 0.05 |iF
reflected
into the primary
6
(b) Primary tank network: parallel resonance
Figure 4.3. Equivalent circuit of an HV transformer.
Selection of number of turns is based on:
1. Desired resonant frequency; nowadays, this is 100 kHz rather than the former 20 kHz.
2. Desired output voltage predominantly determines number of turns, and the coupling coefficient and loading
need to be factored in.
3. Ferrite Core material, style, size, gap, most commonly used.
4. Desired C s .
5. Wire Size: Wire thinner than AWG 38 is very difficult to wind. This small wire size causes no problems as
far as overheating is concerned in the secondary, if the power supply purpose is biasing rather than delivery
of current. Potting the secondary further helps by conducting or spreading the heat around. (Thermal con-
ductivity of potting polymers is low, but is higher than for vacuum!)
6. The winding method must accommodate the necessary winding to winding voltage limit, potting accessibil-
ity, and interwinding capacitance control. Thus, winding method should be loose, such as “lattice” winding
or “bank” winding even if it decreases coupling. Layer separation may be incorporated using a porous glass
fiber or polyester mat material.
7. Impregnation material considerations — see discussion below.
8. Partial Discharge (Corona) test considerations.
9. HV breakdown considerations determine whether pot core or C-cores are used.
Spacing from HV winding to other windings and to the core follow the usual guidelines, that is, there should be no
more than 50 V/mil average field strength at most.
Chapter 4: High Voltage Parts
4-9
High Voltage Power Supply Design Guide for Space
It is also highly desirable to limit the voltage between adjacent layers of windings in the secondary to 200 V, so that
even if the impregnation with potting has an inadvertent void, there will be no corona in the void if the voltage across
it is below the Paschen minimum voltage.
B. Design Sequence
Design will involve interactive processes:
1. Select approximate core size and material. Use Ferrite material below the Magnetic Field Strength, B, of 1000
gauss.
2. Determine primary turns from Faraday’s law.
3. Determine secondary turns from turns ratio and wire size.
4. Estimate size of secondary winding based on items 4-9 on previous page. Check fit on selected core.
5. Determine gap and the resonance capacitor. Build a transformer and try it. The larger the secondary windings
volume, the lower the V s output voltage. Try to keep C s low. Adjust L M by transformer gap.
Design Considerations to keep in mind are:
1. Ferrite material is almost always used because of good electrical performance and available with rounded
smooth surfaces compatible with many materials. Ceramic Magnetics Co., NJ, makes custom shapes.
2. Windings are usually installed on a bobbin and inserted onto core as an assembly. This method allows for
better control of winding placement. Toroids are not practical. Use C-shaped ones or pot cores.
3. Wire sizes smaller than #38 are difficult to wind.
4. Generally, C s capacitances can be minimized but will form the dominant capacitance in tank network. C s
can be minimized by:
a. Selection of impregnation material. Potting compound raises C s by factor of K,
b. Winding method:
• Layer winding produces highest C s . Layer spacing controls C s .
• Lattice, bank, and section winding techniques lower C s .
• Generally, techniques that lower C s also produce reduced field stress.
C. Magnet Wire Selection
Procure to Federal Spec. JW1177A.
Insulation
1. It must be compatible with potting material:
a. Polyester insulation generally acceptable. Low voltage use.
b. Polyimide (ML) insulation provides greatest breakdown strength, however, extremely difficult to strip.
c. Some wire comes with lubricant overcoat. Will result in extremely poor adhesion. Avoid this.
2. For operation at 100 kHz; dissipation factor will be a consideration. D.F. of polyurethanes is too high. Poly-
esters and epoxies have lower D.F.s.
3. Lattice, bank windings require friction surface over insulation. (Phelps/Dodge.) This also has high Breakdown
(B.D.) strength.
Wire Termination
1. Free ends of magnet wire can serve as connection leads. Difficult to prevent small wires from breaking dur-
ing handling. This technique is however used by Code 563 at GSFC.
4-10
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
2. Ends of magnet wire connected to terminals on assembly header. Mechanical stresses can produce wire
breakage.
3. Ends of magnet wire connected to hookup wire and external connections made to hookup wire. Difficult to
dress hookup wire.
D. High Voltage Winding Encapsulation of Transformers
Non-Potted
1. Kapton tape used as a barrier between layers approximately every 300 V. Breakdown field of Kapton tape
is 100 V/mil.
2. Leads entering transformer should be shielded with Kapton tape on HV winding and on adjacent low voltage
windings. (Pot core transformers.)
Potted
1. Porous barrier material must be used. Kapton, mylar, etc., must be avoided. Fiberglass matting, polyester
matting are used. Porous material between each layer aides in wicking in the encapsulant.
2. Potting may be performed on bobbin winding assembly or on complete transformer assembly.
a. Potting of bobbin assembly first, allows for better inspection.
b. Potting of core material must be done with semi rigid material to avoid magnetostriction. So do not
pot the core — or use polyurethane, which is semi-rigid.
c. Complete penetration of EN-11 will occur into layer-wound winding, using #38 AWG maximum wire
and applying matting between alternate layers.
d. Lattice wound coils have shown complete impregnation without use of matting material. More inves-
tigation needs to be done.
e. Potting makes for a more rugged construction, however, some of the properties are contradictory, such as
in “b” above and yet needing the lower D.F. of epoxies; then only pot the HV winding, not the whole trans-
former.
E. Qualification Testing
1. MIL STD 981 requirements.
2. Vacuum life tests.
3. Corona testing.
a. With the Biddle system, it can only be performed interwinding, using DC voltages, if transformer
layout provides separation between complete high voltage winding and low voltage windings/core.
b. With a non-standardized, high pass filter, corona test system, it can be performed in a semi-qualitative
manner under the transformer’s own input power. See Appendix IV, in Chapter 5.
F. Dielectric Stress Design
If a transformer has to have lower voltage outputs “floating” on an already high voltage transformer, then resin coatings
containing powdered carbon can be applied to the proper surfaces of the transformer ferrite core to reduce dielectric
stress. See Ref. [8], This is analogous to corona strand shielding in HV cables.
Chapter 4 : High Voltage Parts
4-11
High Voltage Power Supply Design Guide for Space
III. HV Diodes
Glass-encased diodes should be used rather than plastic molded. The glass encasement should be bubble- or void-free.
This is best achieved with so-called Metoxilyte glass emulsion melted down onto the diodes, not a glass sleeve melted
down over connecting ribbon. (The latter is done by Unitrode and by Phillips.)
Table 4.3. High voltage diodes guide (IN5184, IN649, and IN4586 have also been used).
Manuf.
Type
Rating
Spec.
Semtech
HF60A
6 kV
GSFC 73-15077
Semtech
HF75
7.5 kV
Semtech
HF40
4.0 kV
Semtech
HF15
1.5 kV
VMI (Voltage
Multiplier Inc.)
X150UFG
15 kV
Solid State
Devices, Inc.
SHM 15 UF
through 140 UF
1.5 kV through
14 kV
Microsemi-
conductor
MC002
The reverse recovery time of the diodes should be less than 200 ns. The junction capacitance should be 1 to 2 pF
because the output ripple on a HV multiplier is a function of that. Since the physical size of a 6 kV diode is not much
larger than a 2.5 kV diode, one can well order the 6 kV unit and derate by more than 50% for greater reliability.
For special applications, such as in Optocoupler, diodes with higher than usual reverse leakage currents and slower
ones with 1 ps reverse recovery time are desirable (these are the 15 kV units by VMI). For a while, the Hewlett Pack-
ard Company was well known for advice in the Optocoupler field, but now GSFC has used Micropac Industries, Inc.
units, such as the 6N140 Optocoupler 66012; also the Optek Co. is expert in this area.
IV. High Voltage Transistors
High Voltage Transistors rarely make up, or are found in, high voltage assemblies because these assemblies can be
designed by use of high voltage diodes which produce fewer problems. Nevertheless, there are some important uses
for high voltage transistors. By definition, we call them high voltage transistors if either BV cbo or BV CE0 are greater
than 150 V. (CBO or CEO means collector-base open/collector emitter-open, respectively.)
4-12
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 4.4. HV transistors guide.
Type
Package
bv ce/B0
Specific Unit
NPN
TO-5
300 V
2N3501
2N5662
NPN
T0-3
>300 V
2N5097
PNP
TO-5
300 V
2N5096
Power MOSFET*
TO-5
1000 V
TO-254 AA
1000 V
IRFMG 40
TO-257 AA
1000 V
IRHY7G 30CMSE
*Metal-Oxide Semiconductor Field-Effect Transistor
The earlier power MOSFETs were sensitive to high energy radiation, that is, cosmic rays, solar proton flux, etc. One
can now get radiation hardened ones, for instance, from International IR Rectifier Co. Below follow some outlines of
circuit schematics where high voltage transistors are used. These circuits are in Oscillator Switching, Shunt Regula-
tors and Commandable High Voltage Switches (Figures 4.4-4.7).
(1) High Voltage Oscillator Switching:
Use
2N3501
or 2N5662
Spec. Mil Std 975
}
Current:
0.5 to 1 A
(2) or Fly-back Design:
Figure 4.4. High voltage oscillator switching.
Chapter 4: High Voltage Parts
4-13
High Voltage Power Supply Design Guide for Space
to less
than 1 /rA ki
£
Suitable transistors: 2N5097 NPN or 2N5096 PNP
One needs to stack many transistors (10) in series here.
It is better to use optocouplers.
Power MOSFETS are not suitable here because their leakage currents are too high.
Figure 4.5. Shunt regulators.
HVPS
3kV
/_
t
Command
Possibly also use Spark
Gap Tubes, as was
done in EGRET (GRO).
Using Transistors, this Command “Switch” is drawn in greater detail:
1 vA
These transistors can
be purchased from
Solid State Devices
or
Solitron
Because the current to be controlled by the switch is so low (5 /rA), the leakage
current must be less than 1 /vA. Also, the solder pads must be rounded.
Figure 4.6. HV switch, commandable.
4-14
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Vc
HV Side
of
Transformer
(v s -v R )
\7
t
V R
I
Figure 4.7. Command on FIV side of transformer (K. Castell’s supply).
V. High Voltage Optocouplers
Stepping high voltage applications require output voltages to typically cover a voltage range that spans four decades.
Some applications call for live decades. Additionally, the voltages are swept in logarithmic fashion with durations of
3-10 ms. High slew rates and short settling times are needed to minimize the sampling dead time. At the same time,
the trade-off between stepping rates and power consumption needs to be addressed.
Multi-stage bipolar transistor shunt regulators have been employed in the past to produce stepping outputs with fast
“downward” voltage steps, but increasing voltage steps were controlled by long time-constant RC networks.
In the early 1980s, designers started using optocoupler techniques instead of the bipolar shunt regulators. The photo-
conductive characteristic of a reversed biased high voltage silicon diode served as the means to produce a light con-
trolled, variable high voltage impedance. Illuminating the surface of transparent/translucent coated diodes are 890 nm
GaAlAs infrared LEDs, which produce the necessary HV diode leakage to achieve the slew rates better than 3 kV/ms
(100 pF load). Other light sources have also been successfully used.
A stepping supply block diagram is shown in Figure 4.8 using optocouplers:
Chapter 4: High Voltage Parts
4-15
High Voltage Power Supply Design Guide for Space
+28V
INPUT
FILTER
ON
CMD
CLOCK
STEP
SEQ.
LOGIC
D/A
CONV/
REF.
GEN.
+3000V
-3000V
MON
OPTO
COUPLER
DRVR.
-2500 TO
+2500
OUTPUT
~v_
A A
-3000V
-2500 TO
Figure 4.8. A stepper block diagram using optocouplers.
A fixed level, ± 3000 V output high voltage supply provides the inputs for the output regulators. A linear closed loop
control senses the output voltage, compares it to the sweep reference, and adjusts the LED current accordingly.
The optocoupler approach yields the following advantages:
1. Fast slew rates stepping in both directions.
2. The ability to sweep from +high voltage to -high voltage from a single output.
3. Reduced size, weight, and parts count.
4. A simple linear control loop.
Supplies stepping up to 10 kV have been produced using discrete LEDs and high voltage diodes. An assembly with
the proper LED to high voltage diode spacing was designed to provide the necessary dielectric breakdown strength
and meet the slew rate requirements. Voltage Multiplier, Inc. manufactures diodes specifically intended for this ap-
plication. AMPTEK, Inc. manufactures an optocoupler (HV601) for applications up to ±3000 V.
REFERENCES
1. Plante, J.F., and M.J. Sampson, “Evaluation of High Voltage Multilayer Ceramic Capacitors for Space Flight
Applications,” NASA Parts Project Office/Unisys Support Group, Lanham, Maryland, 1993.
2. “Performances Specification, Capacitor, Fixed, Ceramic, Multilayer, High Voltage,” MIL-PRF-49467B, May
2 , 2001 .
3. Johanson Dielectrics, Understanding Chip Capacitors. Technical Bulletin 792, 1974.
4. Reynolds Industries, Inc., High Voltage Connector Catalog and Handbook. 1991.
5. R.S. Bever, “Investigation of Problems Associated with Solid Encapsulation; also Reynolds Connector
Study,” Goddard Space Flight Center, Greenbelt, Maryland, X-711-75-221, September 1975.
6. Sutton, J.F., and J.E. Stern, “Spacecraft High Voltage Power Supply Construction,” NASA TN-D-7948, God-
dard Space Flight Center, Greenbelt, Maryland, April 1975.
7. Ruitberg, A.P., and J.A. Gillis, “Design Techniques for Miniaturized Spacecraft High Voltage Power Sup-
plies,” Proc. Power Con, 6, Power Concepts, Inc., 1976.
8. Ruitberg, A.P., “Dielectric Stress Design,” Proc. Power Con. 9, Power Concepts, Inc., 1982.
4-16
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
APPENDICES TO CHAPTER 4
Appendix I to Chapter 4
Evaluation of High Voltage Multilayer Ceramic Capacitors for Space Flight Applications, by J.F. Plante and M. J.
Sampson, NASA Parts Project Office/GSFC/Unisys Corp., Lanham, MD, 1993.
The great detail shows the care applied to HV multilayer capacitor evaluation.
Appendix II to Chapter 4
Potting Procedures for the Cassini Plasma Spectrometer (CAPS) HVPS TRANSFORMER & FILTERS , by Suong Le,
Unisys Government Systems MA-1995-169; Oct. 26, 1995.
This gives directions for a difficult potting procedure.
Chapter 4: High Voltage Parts
4-17
High Voltage Power Supply Design Guide for Space
Appendix I of Chapter 4
EVALUATION OF HIGH VOLTAGE MULTILAYER CERAMIC CAPACITORS
FOR SPACE FLIGHT APPLICATIONS (-1993)
Jeannette F. Plante and Michael J. Sampson
NASA Parts Project Office/ Unisys Support Group
ABSTRACT
NASA designers frequently use high voltage ceramic disc capacitors for filtering and smoothing, in power supplies.
Driven by the need for increased volumetric efficiency, NASA has evaluated high voltage multilayer ceramic capaci-
tors (HVMLCs) for spaceflight use.
Electrical and environmental tests were performed to show part reliability with respect to the space flight environment.
AC and DC partial discharge tests were included in the test plan to investigate their effectiveness as nondestructive
screens for potential life test failures.
It was found that an unexpected mechanism stimulated the majority of the life test failures. A much more effective
screening method can now be developed.
BACKGROUND
There is a need within NASA for high voltage capacitors that have a greater volumetric efficiency (capacitance per
unit volume) than the single layer ceramic disc capacitors currently in use. High voltage multilayer ceramic (HVMLC)
styles are now available that offer this size advantage. The military specification MIL-C-49467 covers this part type,
however there are no qualified sources available.
The susceptibility of thin and brittle layers of dielectric to voltage breakdown is the major reliability issue when
considering HVMLCs.
Also the large physical size of these devices makes them vulnerable to damage due to thermal cycling. Although
encapsulated, these parts are not hermetically sealed, therefore moisture contamination is also a concern.
DC partial discharge testing has traditionally been used by NASA as a nondestructive screen for internal defects in
high voltage ceramic disc capacitors. A partial discharge is an electric pulse detected within the dielectric of a ca-
pacitor related to ionization activity. This ionization occurs in regions having lower resistivity than the surrounding
dielectric such as voids and other defects. When the breakdown voltage of the defect is reached a transient current
flows through the fault site and the voltage drop is detected as a pulse measured in picocoulombs. The magnitudes of
the discharges recorded during DC partial discharge testing are a function of the dielectric constant, the ferroelectric
nature of the dielectric, and the relationship between the physical geometries of the defect and their orientations to
the applied electric field.
When an AC signal is applied to these parts, internal partial discharge activity increases drastically and hot spots can
develop at defect sites. Therefore the AC partial discharge test is viewed by NASA as a destructive test when applied
to a part intended for DC use. However, the AC partial discharge test is much more attractive for standard use than
the DC test because it is less time consuming to perform, the equipment is less costly than the DC test equipment
and the data is less open to varied interpretation than that produced by the DC test. If the AC test could be shown
4-18
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
to be nondestructive, the test method could be standardized for incorporation into military and NASA procurement
documents.
OBJECTIVES
The primary objective of the test program was to evaluate a matrix of HVMLC capacitors in order to determine if
the technology exists to provide reliable parts for spaceflight use. Thermal cycling, low voltage steady state humidity
testing and vacuum testing were included in the test plan to address specific space flight reliability concerns.
A second objective was to investigate the effectiveness of DC and AC Partial Discharge (PD) testing as non-destruc-
tive screens for potential life failures.
TEST PLAN
Table 1 [4.A.I.1] shows the test plan and Table 2 [4.A.I.2] lists the test methods used. Each set was submitted to exactly
the same test routine.
Table 1 [4.A.I.1], Test Plan
Screening Tests: (36 pieces/Lot)
Initial Electrical Tests
Construction Analysis
DC Partial Discharge
AC Partial Discharge
Temperature Cycling & Voltage Conditioning
DC Partial Discharge
AC Partial Discharge
Environmental Tests
100 Thermal Cycles
85°C/85% Relative Humidity, Low Voltage
DC Partial Discharge
Barometric Pressure/DC Partial Discharge
Life Test: 1000 hours
DC Partial Discharge Tested
AC Partial Discharge Tested
100 Thermal Cycles tested
Barometric Pressure Tested
Control parts
Chapter 4: High Voltage Parts
4-19
High Voltage Power Supply Design Guide for Space
Table 2 [4.A.I.2]. Test methods for evaluation.
Test
Standard/Specification
Test Condition
Dielectric
Withstanding
Voltage
MIL-STD-202
- Dielectric
Ramp voltage, 120% rated voltage between the leads.
Fail: current greater than or equal to 1 |xA.
- Body
Ramp to 500 V DC, applied between leads and metal
tape wrapped around body.
Capacitance
and
Dissipation
Factor
MIL-STD-202
For X7R dielectric: 1 V at 1 kHz
For NPO dielectric: 1 V at 1 kHz except for
C < 1000 pF then 1 MHz
Fail: Cap: outside of tolerance,
DF: > 2.5% for X7R and > 0.2% for NPO.
Insulation
Resistance
MIL-STD-202
500 V DC applied, Fail: below limit of 100 GO
at 25°C and 10 GO at 85°C, except for one set
where C > 10,000 then 66 GO and
6.6 GO respectively.
Temperature
Cycling
MIL-STD-202
5 cycles from -55°C to +85°C for screening, 100 cycles
for environmental test.
Voltage
Conditioning
MIL-C-49467
Ramp voltage, except use 120% DC rated voltage,
96 hours minimum.
DC Partial
Discharge
Adapted from
ASTMD 1868-81 &
IEEE STD 454-1973
Stepwise ramps and plateau sequence: acquiring for
60 s, the pulse and total charge sum data from,
0 to half rated voltage, half rated voltage to full rated
voltage, at the quiescent plateau and from rated
voltage to 0.
AC Partial
Discharge
Adapted from
ASTMD 1868-81 &
IEEE STD 454-1973
Ramped 60 Hz AC voltage until 100 pC pulses are
detected. Dwell at corona inception voltage maximum
of 5 s. No measurement above 70% of the
DC rated voltage (equivalent in rms volts).
85°C/85%
Relative
Humidity
Low Voltage
MIL-C-123A
Electrical measurements do not exceed limits above.
Capacitance delta not to exceed 10%.
Barometric
Pressure/
DC Partial
Discharge
Adapted from
MIL-STD-202
DC Partial Discharge measurements taken as noted
above at atmospheric pressure in the vacuum chamber.
Pressure reduced to 6(1 O' 6 ) torr and held for 2 h.
DC Partial Discharge measurements taken while parts
are under vacuum.
Life Test
See voltage conditioning above, except use 120% DC
rated voltage. DWV, capacitance, DF and IR
measurements after 240 and 1000 h.
4-20
Chapter 4: High Voltage Parts
PARTS DESCRIPTION
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
NASA procured custom parts based on commercially available designs from three manufacturers. The external di-
mensions complied with EIA-IS-37. All of the parts were radial leaded.
Two dielectric types were selected, NPO (low dielectric constant) and X7R (higher). Three kilovolt and five kilovolt
ratings were specified. The thickness of the dielectric in the active area was restricted to 100 V/mil maximum for parts
made with the X7R dielectric and 200 V/mil maximum for those made with the NPO dielectric. The manufactur-
ers were also asked to provide a lot consisting of their 5 kV, X7R part with the highest capacitance available, in the
standard EIA-IS-37 case size. No V/mil restriction was placed on this last lot.
Capacitance values were not specified (other than the rather loose requirement for the final lot) and no screening tests
were required of the manufacturers. See Table 3 [4.A.I.3] for the final part matrix and some of the characteristics of the
parts that were received. Manufacturer 1 (MFR1) used epoxy backfilled boxes for chip encapsulation. Manufacturer
2 (MFR 2) used urethane backfilled boxes and Manufacturer 3 (MFR 3) used a fluidized bed epoxy coating. MFR 1
supplied two lots, MFR 2 supplied three lots and MFR 3 supplied five lots.
Table 3 [4.A.I.3], Parts matrix.
Set 2: 3 kV, X7R, < 100 V/mil, DF < 2.5%
Number of
Dielectric
Minimum
V/mil
Chip
Capacitance
DF
Plate Design
Dielectric
Thickness
End Margin
(Between
Manufacturer
Nominal
Nominal
2/
Layers 1/
1/
Thickness
Plates)
Height
Width
Thickness
1
4800 pF
1.0%
3 parallel
3
30 mil
34 mil
100
435 mil
432 mil
116 mil
3
3800 pF
1.4%
4 series x 20 parallel
20
7 mil
20 mil
100
319 mil
356 mil
190 mil
Set 4: 5 kV, X7R, < 100 V/mil, DF < 2.5%
2
6600 pF
0.8%
2 series x 8 parallel
8
23 mil
59 mil
100
740 mil
653 mil
276 mil
3
8000 pF
1.4%
4 series x 12 parallel
12
13 mil
30 mil
100
603 mil
771 mil
212 mil
Set 5: 3 kV, NPO, < 200 V/mil, DF < 0.2%
1
400 pF
0.01%
3 parallel
3
16 mil
42 mil
200
434 mil
436 mil
101 mil
3
310 pF
>0.2%
3 series x 20 parallel
20
7 mil
19 mil
200
214 mil
309 mil
183 mil
Set 7: 5 kV, NPO, < 200 V/mil, DF < 0.2%
2
1600 pF
>0.2%
8 parallel
8
22 mil
56 mil
200
732 mil
674 mil
218 mil
3
730 pF
0.02%
4 series x 22 parallel
22
6 mil
28 mil
200
615 mil
730 mil
181 mil
Set 9: 5 kV, X7R, No V/mil Limit, DF < 2.5%, Highest C Possible
2
14000 pF
1.4%
2 series x 10 parallel
10
18 mil
24 mil
140
700 mil
598 mil
248 mil
3
10800 pF
1.2%
4 series x 14 parallel
14
10.5 mil
30 mil
120
600 mil
733 mil
190 mil
1/ Dielectric layers counted at thinnest points
2/ See Figures below
4 Series x 8 parallel
[Different plate designs in high voltage ceramic multilayer capacitors.]
Chapter 4: High Voltage Parts
4-21
TEST RESULTS
High Voltage Power Supply Design Guide for Space
Material Tests
Per ASTM-E-595, “Standard Test Method for Total Mass Loss and Collected Volatile Condensable Materials From
Outgassing in a Vacuum Environment,” outgassing testing was performed on prepared samples of the three material
combinations used for element encapsulation. Collected Volatile Condensible Materials (CVCM) data for the epoxy
exceeded the limit by 0.03%. The urethane and proprietary coating passed the outgassing requirement (Total Mass
Loss < 1.0%, CVCM <0.1%).
Screening Tests
All of the test parts were screened using standard electrical test (See Table 2 [4.A.I.2]). Insulation resistance (IR)
testing which was performed at 25°C and 85°C, using a 500 V DC bias for 2 minutes showed the only initial failures.
Large numbers of MFR 3’s parts (three out of live lots) showed IR failures at 85°C. The problem was traced to moisture
absorbed by the coating. Passing IR values (>100 GQ) were measured after these parts were baked for 12 hours at
125°C. DC and AC partial discharge testing was performed on two subgroups of each lot (10 pieces each) following
initial electrical characterization. Results of the partial discharge testing will be discussed later.
Table 4 [4.A.I.4]. Summary of screening failures.
36 pieces tested
in each lot
Lot and
Manufacturer
Incoming
No.
Failures
Type
Post Voltage
Conditioning
Failures
No. Type
Capacitance
Change From
Incoming
Nominal (%)
Dissipation
Factor Change
From Incoming
Nominal (%)
Lot A: X7R, 3 kV, <100 V/mil
1
1
Cap
7
DWV
0%
0%
3
0
—
35 1/
CAP
-11%
-0.04%
Lot B: X7R, 5 kV, <100 V/mil
2
0
1
DWV
-4.6%
0%
1
DWV
3
43
IR85C
36 1/
CAP
-7.6%
-0.02%
Lot C: NPO, 3 kV, <200 V/mil
1
1
Cap & DF
2
Cap & DWV
0%
0%
3
37
IR85C
0
—
0%
-0.01%
Lot D: NPO, 5 kV, <200 V/mil
2
2
Cap
2
DWV
-6%
0%
3
8
1R85C
2
DWV
0%
-0.02%
Lot E: X7R, 5 kV, No <100 V/mil limit
2
0
—
0
—
-3.7%
0%
3
0
—
0
—
-9%
0%
1/ Measured below 10% capacitance tolerance
4-22
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
All of the parts (36 pieces) were submitted to the following testing sequence: five thermal cycles (-55°C to 85°C),
DWV tests (dielectric and body test), voltage conditioning (120% DC rated voltage at 85°C, for 100 hours), DWV
(dielectric and body), Capacitance, DF, IR at high and low temperature, and DC partial discharge.
Subgroup II was AC partial discharge tested a second time instead of DC partial discharge tested at this point. See
Table 4 [4.A.I.4] for a summary of the screening failures.
Construction Analysis
Two capacitors from each lot underwent construction analysis. All of the samples were visually examined externally
and then cross-sectioned at 25%, 50% and 75% into the capacitor from the top down toward the two leads.
Table 5 [4.A.I.5] shows the results of this examination. The definition of a rejectable defect is in accordance with the
destructive physical analysis listed in EIA-469-B. See Table 3 [4.A.I.3] for chip dimensions and plate arrangement
information.
Table 5 [4.A.I.5]. Summary of internal defects.
Lot and
MFR
Rejectable
Defect 1/
Lot A:
X7R, 3 kV, <100 V/mil
1
Crack
3
None
Lot B:
X7R, 5 kV, <100 V/mil
2
Delamination
3
None
Lot C:
NPO, 3 kV, <200 V/mil
1
None
3
Delamination
Lot D:
NPO, 5 kV, <200 V/mil
2
None
3
Delamination
Lot E:
X7R, 5 kV, No V/mil Limit
2
Crack
4
None
The crack seen in one of MFR l’s parts was probably due to thermal shock during the lead attach process. The same
mechanism appeared to have produced nonrejectable cracks in the cover plate, outside of the active area. These nonre-
jectable cracks were found in both lots received from MFR 1. MFR 2’s parts showed delaminations probably induced
during ink binder burn out. In one sample, a large crack at the end termination was evidence of severe thermal shock
during that burn out stage. MFR 3’s parts showed cracks in the cover plate due to the encapsulation process. Several
parts showed severe delaminations. Lamination lines were obvious as well.
Chapter 4: High Voltage Parts
4-23
High Voltage Power Supply Design Guide for Space
Environmental Tests
Subgroups were subjected to either one hundred thermal cycles (-55°C to 85°C) or the 85/85 test (85°C, 85% relative
humidity, 1.3 V DC bias, 240 hours). DWV, Capacitance, DF, IR at high and low temperature and DC partial discharge
measurements were made after the temperature cycles. The same applied after the 85/85 test with the exception of DWV
testing. Table 6 [4.A.I.6] lists the failures found after thermal cycling. Cracks could be seen in the coating on several
of MFR 3’s larger parts (lots B, D and E) after thermal cycling even though they did not show electrical failure. All of
MFR l’s and MFR 3’s lots showed failing IR measurements immediately after the 85/85 test (within 24 hours). MFR
2’s lot E parts only (X7R, 5 kV, 140 V/mil), showed this type of failure. All of these failing measurements recovered
after oven baking for 12 hours (85°C).
Table 6 [4.A.I.6]. Summary of Thermal Cycling failure (100 cycles).
Lot & Mfr.
Failures (10 pieces tested)
No.
Type
Lot C:
NPO,
5 kV, 200 V/mil
3
2
DF& IR at 85°C
Lot E:
X7R,
5 kV, No V/mil limit
2
5
C
The balance of MFR 2’s pieces (lot B and lot D) did not show any electrical failure after the 85/85 test.
A final subgroup consisted of parts that had seen one hundred thermal cycles, parts that had seen 85/85 testing and
one part that had only seen screening tests. DC partial discharge measurements were taken on these parts after two
hours in vacuum (6 x 10 6 torr) in an attempt to identify voids in the potting that could have been bridging the end
terminations. No significant increase in partial discharges was detected compared to measurements taken at atmo-
spheric pressure.
Life Tests
Life testing was performed with all of the parts completely submerged in Fluorinert dielectric fluid for 1000 hours.
The parts were biased with 120% of the DC rated voltage and the temperature was maintained at 85°C. The start up
procedure was as follows: Submerge fixture (5 boards, 36 pieces in parallel per board) into Fluorinert bath (room
temperature). Slowly (25 seconds) raise voltage on power supply to 120% DC rated voltage. Confirm voltage at sev-
eral locations on the board. Set bath temperature to 85°C. Temperature rise time from room temperature to 85°C was
approximately 1.5 hours.
The dielectric fluid was constantly circulated in the bath for an even temperature distribution. This start up procedure
provided protection against uneven temperature exposure or system induced thermal shock. Currents in excess of 3 m A
automatically shut down the power supply. IR at high and low temperature, DWV, Capacitance and DF measurements
were made following 240 and 1000 hours of life testing.
4-24
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 7 [4.A.I.7]. Summary of life failures.
Lot and Internal Plate
No. of
Hour of
Type
Failure Analysis
Manufacturer Design
Failures
Failure
1/
Summary
Lot A: X7R, 3 kV,<100 V/mil
1 3 Parallel
0
—
—
—
3 4 Series x 21 Parallel
0
—
—
—
Lot B: X7R, 5 kV,<100 V/mil
2 3 Series x 9 Parallel
0
—
—
—
3 4 Series x 13 Parallel
1
262
Restart
Crack through 3 layers
and the cover plate
Lot C: NPO, 3 kV, <200 V/mil
1 3 Parallel
3
0
Restart
All three of these
0
Restart
failures showed a
301
Restart
crack through cover
3 3 Series x 21 Parallel
0
plate and breakdown
from the first
metal plate to the
end termination
Lot D: NPO, 5 kV, <200 V/mil
2 9 Parallel
1
643
Restart
No defect found
3 4 Series x23 Parallel
1
623
Restart
Cracks perpendicular
to and bridging metal
plates, delaminations
Lot E: X7R, 5 kV, No V/mil limit
2 3 Series x 1 1 Parallel
4
364
Restart
Crack in ceramic
parallel to and next to
the end tennination
671
Restart
Crack in the ceramic
parallel to and next to
the end termination
215
During
Crack at the top margin
Test
through to the metal
plates
416
During
Crack through cover
Test
plate through to first
metal plate
3 4 Series x 15 Parallel
1
376
Restart
Cracks perpendicular
to and bridging metal
plates, delaminations
1/Restart: Failure during a restart of the burn in system, temperature > 60C, 120% DC rated voltage applied.
During Test: Failure well after burn in stabilization period (at 85°C with voltage applied).
Deltas (capacitance and DF) were calculated using electrical data taken just before the life test, and data taken fol-
lowing 240 hours of life testing for subgroups III, IV and V and data taken following 1000 hours of life testing for
subgroups I and II. All of the deltas calculated showed values of less than 2%.
A matrix showing life test failures and associated lot characteristics is given in Table 7 [4.A.I.7]. These failures were
breakdowns that caused the power supply to shut down. While compiling Table 7 [4.A.I.7], it was noticed that nine
out of eleven of the failures occurred during a restart of the system. These “start up” failures were found in all lots
Chapter 4: High Voltage Parts
4-25
High Voltage Power Supply Design Guide for Space
that had a life failure. Failure analyses showed failure sites that corresponded to defect areas found during construc-
tion analysis (Table 5 [4.A.I.5]).
Table 8 [4.A.I.8]. Summary of partial discharge results.
AC Partial
Discharge
Rating 1/
DC Partial
Discharge
Rating 1/
Defects
Found
During
Construction
Analysis
Life
Failures
No.
Lot
Description
A
A
None
3
Lot C, NPO, 3 kv, 200 V/mil
MFR 1,400 pF
3 Parallel
B
B
None
0
Lot A, X7R, 3 kV, 100 V/ mil
MFR 3, 3800 pF
4 Series x 20 Parallel
C
D
None
1
Lot E, X7R, 5 kV, no V/mil restriction
MFR 3, 10,800 pF
4 Series x 14 Parallel
D
C
None
1
Lot D, NPO, 5 kV, 200 V/mil
MFR 2, 1600 pF
8 Parallel
E
V
Crack
0
Lot A, X7R, 3 kV, 100 V/mil
MFR 1, 4800 pF
3 Parallel
V
E
None
1
Lot B, X7R, 5 kV, 100 V/mil
MFR 3, 8000 pF
4 Series x 12 Parallel
W
X
Delamination
1
Lot D, NPO, 5 kV, 200 V/mil
MFR 3, 730 pF
4 Series x 22 Parallel
X
w
Delamination
0
Lot C, NPO, 3 kV, 200 V/mil
MFR 3, 310 pF
3 Series x 20 Parallel
Y
Y
Delamination
0
Lot B, X7R, 5 kV, 100 V/mil
MFR 2, 6600 pF
2 Series x 8 Parallel
Z
Z
Crack
4
Lot E, X7R, 5 kV, no V/milrestriction
MFR 2, 14,000 pF
2 Series x 1 0 Parallel
1/ABCDEVWXYZ
Best— ^Moderate— s*— »Worst
Table 9 [4.A.I.9], Exerpts from partial discharge data [both DC and AC],
DC PARTIAL DISCHARGE
AC PARTIAL DISCHARGE
RATING:
: A
RATING: A
RATING: Z
Serial
0 to V/2
V/2 to V
Hold at V
VtoO
Serial
CIV for
CIV as
Serial
CIV for
CIV as
Number
# of
max
sum of
#of
max
sum of
#of
max sum of
#of
max
sum of
Number
100 pC
% of DC
Number
100 pC
% of DC
pulses pulse pulses
pulses pulse pulses
pulses
pulse pulses
pulses
pulse pulses
rated V
rated V
1569
0
0
0
0
0
0
0
0
0
0
0
0
1590
>1.5
>71.4
2990
0.7
20.0
1570
0
0
0
0
0
0
0
0
0
0
0
0
1591
>1.5
>71.4
2991
0.6
17.1
1571
0
0
0
0
0
0
0
0
0
0
0
0
1592
>1.5
>71.4
2992
0.6
17.1
1572
0
0
0
1
19
19
0
0
0
4
85
103
1593
>1.5
>71.4
2993
0.2
5.71
1573
0
0
0
0
0
0
0
0
0
0
0
0
1595
>1.5
>71.4
2994
0.6
17.1
RATING:
1596
>1.5
>71.4
2995
0.6
17.1
: Z
2996
0.6
1597
>1.5
>71.4
17.1
2969
290
1008
39779
517
2750
157198
7
276
943
622
944
53156
1598
>1.5
>71.4
2997
0.4
11.4
2970
221
462
15839
427
2510
112911
15
386
956
296
737
31760
1599
>1.5
>71.4
2998
0.6
17.1
2971
196
1012
27862
268
1506
81891
11
281
877
729
1013
67688
2999
0.6
17.1
2972
233
980
35634
358
2058
123353
8
191
481
497
992
69760
2973
170
978
28287
229
2108
70727
2
54
94
556
442
27750
PARTIAL DISCHARGE DATA
DC:
The DC Partial Discharge test method consisted of measuring charge pulses in picocoulombs while: ramping the volt-
age to half rated voltage, then to full rated voltage, resting at full rated (quiescent) voltage and then ramping to zero
volts. After each of these four measurement segments, analytical software then calculated the following data: the
total number of pulses detected, the value of the largest pulse detected, and the total charge detected. These three data
points taken at each of the four measurement segments periods formed the DC partial discharge signature.
As discussed earlier, the typical DC partial discharge signature for a part type is dependant on material character-
istics and internal electrical design. Therefore, only that data originating from lots that have like design and process
characteristics can be compared. DC partial discharge data is typically used in two ways to screen parts: to reject
4-26
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
individual parts that have discharge sums that are magnitudes larger than the other parts in the lot or, to reject whole
lots which have large partial discharges occurring throughout the four measurement segments.
AC:
The AC partial discharge test consists of the application of a 60 Hz sinusoidal signal, ramped up in voltage until the
Corona Inception Voltage (CIV) for a 100 pC pulse is reached. The 100 pC value is a convenient test standard frequently
used by the electrical insulation industry. A maximum voltage limit of 70% of the rated rms voltage was imposed to
limit unnecessary damage to the part while allowing sufficient measurement range. The highest test voltage was also
maintained for a maximum of 5 seconds to limit [develop damaging] potential damage. AC partial discharge data is
typically used as a screening test to accept parts that do not [have] corona within the operating voltage range. Typi-
cally, NASA imposes a 60% derating factor.
Twenty pieces from each lot were screened, partial discharge tested and life tested to show any damage due to AC or
DC partial discharge testing (10 pieces were DC tested only and 10 pieces were AC tested only).
All ten lots were rated based on the DC and AC partial discharge data. Table 8 [4.A.I.8] shows examples of the best
(rated A) and the worst (rated Z) DC partial discharge signatures and AC partial discharge data. Table 8 [4.A.I.8]
shows the results of this rating exercise. [Also Table 9 [4.A.I.9] shows partial discharge data, both AC and DC.]
Deltas calculated between incoming electrical data and electrical data taken after life test, show no parametric degrada-
tion even after the two exposures to AC partial discharge testing. Only one of the parts that had AC partial discharge
testing failed on life test. This failure mechanism was not associated with AC voltage induced stress.
CONCLUSIONS
HVMLC capacitors continued to show stable electrical parameters during the application of a wide range of dc volt-
ages and environmental stresses. The electrical and life test data, to date, do not show that the 100 V/mil field stress
limit provides improved reliability versus the manufacturer’s chosen design (120 and 140 V/mil). The life test failures
reflected process control related defects and did not reflect a relationship to the V/mil characteristic.
Process related defects dominated the construction analysis findings. Comparison of the construction analyses and
failure analyses showed that the defects were common by manufacturer. Defects found in one manufacture’s parts
during construction analysis were identical to defects found in a different set of his parts during failure analysis. It
is believed that the technology is available to produce HVMLC capacitors, if the manufacturing processes are well
controlled. This evaluation showed that none of the manufacturers involved appeared to have had adequate control
over their processes.
Considering their size and the sensitivity of ceramics to thermal shock, the parts performed well after 100 thermal
cycles from -55°C to 85°C. The majority of the thermal shock failures found were attributed to pre-existing cracks
in the ceramic.
The urethane potting material was effective in protecting the capacitor from moisture penetration. The epoxy fluidized
bed coating did not provide sufficient protection from moisture or thermal stress.
The barometric pressure test was not conclusive. A vacuum potting combined with a post encapsulation DPA will
provide a more efficient and practical defect detection tool.
The failures revealed during the life test correlate well with the construction analysis findings. The majority of these
failures occurred during the power up period while the temperature was increasing (“start up” failure). This indicates
Chapter 4 : High Voltage Parts
4-27
High Voltage Power Supply Design Guide for Space
that a steady state life test is not the best way to detect flawed lots. A better test would involve temperature cycling pos-
sibly combined with voltage switching. The most effective test method remains to be developed. It can be anticipated
that this test method would form the basis for an effective screen (burn in) as well as an appropriate life test.
Although partial discharge testing effectively identifies flawed lots, it is unable to identify some kinds of critical
defects and does not identify lot quality more effectively than sample DPA. Thus, even though it has been shown
that AC partial discharge is nondestructive, it is not recommended that it be used as a specification requirement.
However, the AC partial discharge test could be used as a customer imposed screen when selecting parts for use in
critical applications.
RECOMMENDA TION
It will be recommended that the military specification MIL-C-49467 be upgraded to a Class S specification which
contains the following key features: 100 thermal cycles from -55°C to 85°C; Vacuum potting of the capacitors and
material outgassing limits; Sample DPAs before termination and after encapsulation; An improved screening and life
test (method to be developed).
It is further recommended that the AC partial discharge requirement be removed from the specification.
[AUTHORS’ NOTE to Appendix i, Chapter 4: The authors of this HV Guide disagree with this recommendation that
the AC partial discharge requirement be removed from the specification.]
REFERENCES
Capozzi, V.F., Multilayer Ceramic Capacitor Materials and Manufacture, 1975.
Bever, R.S., “Ramp Technique for DC Partial Discharge,” IEEE Trans. Electrical Insulation, Vol. EI-20 No. 1,
February 1985.
DeMatos, H.V., and C.R. Koripella, “Crack Initiation and Propagation in MLC Chips Subjected to Thermal
Stresses,” 1988.
Alvarez, R., “An Overview of Insulation Stress Testing-Part 2,” Electrical Manufacturing, November 1987.
Ropiak, S., “Characteristics of Low Voltage Ceramic Capacitor Failures,” 1982.
“Non-Destructive Testing of Multilayer Ceramic Capacitors,” Kemet Engineering Bulletin, April 1980.
4-28
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Appendix II of Chapter 4
POTTING PROCEDURES FOR THE CASSINI PLASMA
SPECTROMETER (CAPS)
HVPS TRANSFORMERS AND FILTERS
BY SUONG LE / UNISYS GOVERNMENT SYSTEMS
MA-1995-169
OCTOBER 26, 1995
INTRODUCTION
In support of Code 313, Unisys personnel generated the potting procedures for the CAPS 16 kV high voltage power
supply (HVPS) transformers and filters. The following procedures should be added to the respective transformer and
filter specifications. The following is a log of the processing steps used by Mr. Clatterbuck of Code 313 to process
the protoflight and engineering model components. They are recommended for the flight unit, ft should be noted that
ESD precaution is not a requirement for these procedures.
CAPS TRANSFORMERS
This procedure shall be performed in the Class 10,000 clean room. The Flight Assurance Manager or Quality Control
engineer shall be notified at least twenty-four (24) hours prior to the start of this procedure.
Equipment and Materials:
• Transformer with soldered lead wires and wire -wound core on the Ultem housing from Code 734.3
• Stycast 3050 adhesive with catalyst 11 from Emerson & Cuming
• DC 6-1104 adhesive from Dow Corning
• HM-430 tape from CHR
• 200-proof ethanol from GSA or equivalent
• Reagent-grade heptane from Fisher Scientific or equivalent
• Aluminum foil cup from Fisher Scientific or equivalent
• Ultrasonic cleaner Model No. 8891 DTH from Cole Parmer or equivalent
• Binks No. 15 spray gun from Binks
• Brass cork borer No. 8 from local sources
• Stainless steel spatula from local sources
• Aclar 22C bags from Clean Room Products
• Dry nitrogen gas from local sources
• Vacuum oven with pressure of less than 200 millitorr from Code 313
• Vacuum oven with pressure of less than 400 millitorr from Code 313
• Exhaust fume hood from Code 313
• Soxhlet-extracted swabs from Code 313
• Latex gloves CR-100 from Baxter
• Polyethylene gloves from Fisher Scientific
Chapter 4: High Voltage Parts
4-29
High Voltage Power Supply Design Guide for Space
Cleaning of Hardware
All supporting hardware and materials to be used in this potting procedure shall be cleaned with a 1 to 1 part by
volume of 200-proof ethanol to reagent-grade heptane, then, baked in a convection oven for thirty (30) minutes mini-
mum at 60°C to 65°C.
NOTE: Clean polyethylene or latex gloves shall be worn at all times during processing of the transformer. Only
polyethylene gloves are acceptable for processing with a solvent.
The unpotted transformer shall be cleaned as follows:
1. Soak the transformer in a 1 to 1 part by volume mix ratio of 200-proof ethanol to heptane [ethanol/heptane
solvent] for thirty (30) minutes.
2. Clean the transformer in an ultrasonic cleaner for fifteen (15) minutes at room temperature. The cleaning
solvent shall be the ethanol/heptane solvent.
3. Spray clean for two to three minutes in the ethanol/heptane solvent using the Binks no. 15 spray gun at ap-
proximately 40 psi with clean, dry nitrogen gas.
4. Blow dry the transformer surfaces with clean, dry nitrogen gas.
5. Bake the transformer between 60°C to 65°C for one (1) hour minimum in vacuum at 300 militorr to 400 mili-
torr.
Masking of the Transformer
Masking shall be performed immediately after the cleaning and drying process. The procedure is as follows:
1. Seal electrical feedthrus, around electrical lead wires, with the DC6-1104 silicone adhesive. The silicone
shall be applied only to external surfaces as shown in Figure 1 [4.A.II.1].
2. Air dry the adhesive for one (1) hour minimum at room temperature.
3. Mask the top and bottom of the wire -wound center-core with the Teflon HM-430 tape (Figure 1 [4.A.II.1]).
Use the no. 8 brass cork borer to cut the sealing tape for the through holes.
4. Mask the Ultem transformer housing around the periphery with a 1-inch wide HM-430 tape starting at approxi-
mately 1/8” from the top of the housing (Figure 2 [4.A.II.2]). Masking is to prevent encapsulant spillage during
degassing.
Potting of the Transformer
Potting shall be performed at least one hour after the application of the DC6-1 104 adhesive and immediately after the
masking process. The procedure is as follows:
1. Stir the Stycast 3050 epoxy resin with a stainless steel spatula for four to five minutes or until the resin is
well mixed.
2. Weight out the resin from step 1, then, add in catalyst 11 with an eye dropper. The mix ratio is 100 to 8 parts
by weight of resin to catalyst. The pot life for the Stycast 3050/catalyst 11 adhesive is approximately four (4)
hours.
3. Mix resin and catalyst under the hood for approximately five minutes.
4. Degas the mixed contents for thirty (30) minutes in approximately 200 millitorr vacuum or less. The de-
aerated adhesive is referred to as “encapsulant” hereafter.
NOTE: Resin stirring, mixing, and degassing (Steps 1 through 4) shall be performed under an exhaust fume
hood prior to bringing it into the clean room.
5 . Cast a witness sample of approximately 1 inch in diameter and 14 inch in thickness.
4-30
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
6. Pour the encapsulant slowly into the housing until approximately half full.
7. Degas the encapsulant in the transformer for thirty minutes in approximately 200 millitorr vacuum or less.
8. Remove the transformer from vacuum, then, fill the housing almost full with the encapsulant.
9. Degas the housing from step 8 for thirty (30) minutes in approximately 200 millitorr vacuum or less.
10. Fill the housing with the encapsulant up to the top of the transformer housing.
11. Degas for forty-five (45) minutes in approximately 200 millitorr vacuum or less for complete impregnation
of the resin into the coil and core.
12. Bake the potted transformer and the witness sample of Step 5 in a convection oven at between 83°C to 85°C
for thirty-six (36) hours minimum.
13. Remove the potted transformer from the oven and let it cool to room temperature.
14. Remove the masking tape, DC6-1104 adhesive and clean any adhesive residue from external surfaces of the
casing and wire insulation using a Soxhlet-extracted swab dampened with ethanol.
15. Verify and record the hardness of the witness sample.
16. Place the transformer in a clean Aclar 22C bag for storage or transport.
CAPS FILTERS
This procedure shall be performed in the Class 10,000 clean room. The Flight Assurance Manager or Quality Control
engineer shall be notified at least twenty-four (24) hours prior to the start of this procedure.
Equipment and Materials
• Filter with soldered lead wires and EEE components on the Ultem housing from Code 734.3
• EN-11 part B and EN-4 part A from Conap
• PR-420A/B primer from Product Research
• DC 6-1104 adhesive from Dow Corning
• FIM-430 tape from CF1R
• 200-proof ethanol from GSA or equivalent
• Reagent-grade heptane from Fisher Scientific or equivalent
• Aluminum foil cup from Fisher Scientific or equivalent
• Ultrasonic cleaner Model No. 8891 DTF1 from Cole Parmer or equivalent
• Binks No. 15 spray gun from Binks
• Brass cork borer No. 8 from local sources
• Stainless steel spatula from local sources
• Aclar 22C bags from Clean Room Products
• Dry nitrogen gas from local sources
• Vacuum oven with pressure of less than 200 millitorr from Code 313
• Vacuum oven with pressure of less than 400 millitorr from Code 313
• Exhaust fume hood from Code 313
• Soxhlet-extracted swabs from Code 313
• Latex gloves CR-100 from Baxter
• Polyethylene gloves from Fisher Scientific
Chapter 4: High Voltage Parts
4-31
High Voltage Power Supply Design Guide for Space
Cleaning of Hardware
All supporting hardware and materials to be used in this potting procedure shall be cleaned with a 1 to 1 part by
volume of 200-proof ethanol to reagent-grade heptane, then, baked in a convection oven for thirty (30) minutes mini-
mum at 60°C to 65°C.
NOTE: Clean polyethylene or latex gloves shall be worn at all times during processing of the transformer. Only
polyethylene gloves are acceptable for processing with a solvent.
The unpotted filter shall be cleaned as follows:
1. Soak the filter in a 1 to 1 part by volume mix ratio of 200-proof ethanol to heptane [ethanol/heptane solvent]
for thirty (30) minutes.
2. Clean the filter in an ultrasonic cleaner for fifteen (15) minutes at room temperature (RT). The cleaning
solvent shall be the ethanol/heptane solvent.
3. Spray clean for two to three minutes in the ethanol/heptane solvent using the Binks no. 15 spray gun at ap-
proximately 30 psi with clean, dry nitrogen gas.
4. Blow dry the filter surfaces with clean, dry nitrogen gas.
5. Bake the filter between 60°C to 65°C for one (1) hour minimum in vacuum at 300 millitorr to 400 millitorr.
Sealing and Masking of the Filter
Masking shall be performed immediately after the cleaning process. The procedure is as follows:
1. Seal the areas where the screws are in contact with the external surfaces of the filter housing with the DC6-
1104 silicone adhesive (Figure 1 [4.A.II.1]). Use a syringe for the adhesive application.
2. Air dry the adhesive for one (1) hour minimum at RT.
3. Mask all holes of the filter housing (Figure 1 [4.A.II.1]) with a half-inch wide HM-430 Teflon tape.
Prim ing of the Filter
Primer PR-420A/B can be applied immediately after the DC6-1104 silicone sealing.
1. Stir part A and B well before weighing. The mix ratio shall be 13.7/100 A/B pbw.
2. Let mixed primer stand for fifteen (15) minutes at room temperature before priming all surfaces of the resistors
and capacitor.
Masking of the Filter for Potting
1. Mask the Ultem filter housing completely around the periphery with a half-inch wide HM-430 tape starting
at approximately 1/8” from the top of the housing (Figure 1 [4.A.II.1]). Masking is to prevent encapsulant
spillage during degassing.
2 Reinforce the overlapped area of with the tape. Use a tweezer to force the tape layers in place.
Potting of the Filter
Potting shall be performed at least one hour after the application of the DC6-1104 adhesive and immediately after the
masking process. The procedure is as follows:
1. Stir Conathane EN-11B well for five minutes minimum.
2. Weigh out the resin Conathane EN-4A and curing agent EN-11B in the ratio of 100 to 55 parts by weight of
4-32
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
EN-4A to EN-11B. The pot life for the EN-4A/EN-11B adhesive is eighty (80) minutes maximum at room
temperature.
3. Mix the content under the hood for approximately five minutes.
4. Degas the mixed contents for five (5) minutes in approximately 200 to 600 militorr vacuum. The de-aerated ad-
hesive is referred to as “encapsulant” hereafter.
NOTE: Resin stirring, mixing, and degassing (Steps 1 through 4) shall be performed under an exhaust fume hood
prior to bringing it into the clean room.
5. Cast a witness sample of approximately 1 inch in diameter and 1/2 inch in thickness.
6. Fill the housing with the encapsulant until half full.
7. Degas the adhesive from step 6 for thirty (30) minutes in approximately 200 to 600 millitorr vacuum.
8. Remove the filter from vacuum, then, fill the housing almost full with the encapsulant (1/8” from the top of
the housing).
9. Degas the housing from step 8 for thirty (30) minutes in approximately 200 to 600 millitorr vacuum.
10. Remove the filter from vacuum, then, fill the housing with the encapsulant up to the top of the housing cas-
ing.
11. Let the encapsulant and witness sample in Step 5 gel for 24 hours minimum at room temperature.
12. Bake the potted filter in a convection oven and witness sample at 65°C ±5°C for forty-eight (48) hours mini-
mum.
13. Remove the potted filter from the oven and let it cool to room temperature.
14. Remove the masking tape, DC6-1104 adhesive and clean any potting adhesive residue from the external sur-
faces of the casing and wire insulation using a Soxhlet-extracted swab dampened with the ethanol/heptane
solvent.
15. Verify and record the hardness of the witness sample.
16. Place the potted filter in a clean Aclar 22C bag for storage or transport.
HM-430 tape
Figure 4.A.II.1. Masking of seating.
Chapter 4: High Voltage Parts
4-33
High Voltage Power Supply Design Guide for Space
DC6-1104
DC6-1104
DC6-1104
Primer
PR 420
DC6-1104
DC6-1 104 Adhesive
HM-430 (1/2” wide)
Masking Tape
Around Periphery
oh Housing Casing
Figure 4.A.II.3. Filter masking and sealing.
4-34
Chapter 4: High Voltage Parts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 5. Partial Discharge or Corona Measurement
Introduction
Partial discharge or corona measurements are a well recognized method, under AC impressed voltage, for detecting
the presence of gaseous voids, cracks, delaminations, etc., in electrical insulation. The partial discharges or corona
are small electrical transients through the gaseous spaces, superposed on the main output voltage or current — the
detection of the gaseous voids is thus indirect. Therefore, one must understand the physics of the origin of the partial
discharges and why it is different for AC than for DC impressed voltage. Interpretation is not easy or obvious (Refs.
[1] and [7]).
I. Theory
The equivalent circuit of a void in a dielectric is different for AC than for DC (Figures 5.1 and 5.2).
A. Partial Discharge under AC Applied Voltage
Obviously, 1 kV rms “steady” 60 Hz AC Voltage is not steady. It is up-ramps, down-ramps, voltage reversals, down-
ramps to -1.4 kV peak, etc. Again, the equivalent circuit of a void in a dielectric under AC applied voltage is essentially
two capacitors in series as in Fig. 5.1. The recurrence of internal discharges at the corona inception voltage (CIV) as a
function of applied AC voltage V a is shown in Fig. 5.3. The voltage across the cavity, V„ rises with V a until it reaches
the breakdown voltage across the cavity, and then a flow of free charges occurs through the cavity, making the volt-
age across it almost zero. As the outside applied voltage goes down, then by instantaneous capacitative coupling, the
V c across the cavity reduces below zero to a negative breakdown voltage across the gas, and another current transient
occurs and the cavity voltage again reduces to zero. As the outside applied voltage goes through the negative half-
cycle of AC, there will be two more discharges in the void, as seen in Fig.5.3, for a total of 4 per 1 AC applied cycle,
already at the corona inception voltage; see Refs. [1] and [7],
Summary for AC Partial Discharge
• At corona inception voltage (CIV), already there are theoretically 4 discharges/cycle/1 void
AC:f m = 4 f power at CIV
This means 14,400 PDs/60 s at / power = 60 Hz
(/ PD = frequency of the partial discharges; /power = power frequency)
• Because there are so many discharges on AC, the total effect over time is harmful. Treeing. Catastrophic arc through
the Tree.
• Do not stay at CIV long. Your choice whether CIV at 10, 50, or 100 pC (pC = picocoulomb)
• Do not test above CIV. Your choice whether CIV at 10, 50, or 100 pC.
• Do not use above CIV. Your choice whether CIV at 10, 50, or 100 pC.
• Measure and record CIV. The “trade” uses 100 pC inception.
• The voltage across the cavity depends on the dielectric constant of the insulation in which it is buried. See Fig. 5.4.
Therefore, for a given applied voltage, PDs begin at a lower applied voltage the higher the dielectric constant of
the test sample.
• For acceptance/rejection criteria for capacitors under AC partial discharge testing, see Fig. 5.5.
Chapter 5: Partial Discharge or Corona Measurement
5-1
High Voltage Power Supply Design Guide for Space
Figure 5.1. Equivalent circuit or model for void in dielectric, Ref. [13], The left side of the figure is the
dielectric with void. The right side is the equivalent circuit of void in dielectric for AC partial discharge
testing.
Figure 5.2. Lumped parameter circuit model of a cavity for the DC partial discharge case.
Figure 5.3. AC partial discharge when V a =CIV, Ref. [7],
5-2
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
VOLTAGE ACROSS THE CAVITY OR VOID
Refer to Fig. 5.1 :
(Basic 1 ) C c V c = C b V b = C tota |V ap p|j ed
(Basic 2) V c + V b = V applied
Therefore V c = V appNed C b
+ C c
Assume 1 mil thick pancake void in a 40 mil
thick capacitor, and C = Ke 0 A
t
For X7R ceramic; K = 2000: V c = V applied x 50 = V applied
51
For NPO ceramic; K = 80: V c = V applied x 2 = 2/3 V applied
2 + 1
For Polymer, K = 4. V c = V app n ed x 0.1 =1/11 V app n ed
1 + 0.1
Also, Recharging of Cavity is by INSTANTANEOUS
INDUCTION upon A.C. Applied Voltage
Figure 5.4. AC partial discharge.
ACCEPTANCE/REJECTION CRITERION
(1) [DC RATED VOLTAGE] = RATED RMS VOLTAGE
L 141 J
(2) DERATING RULE for NASA -a DERATE TO 60% V RATED
(3) THEREFORE _ _
CIV MUST BE > |0.60 DC RATED VOLTAGE] VOLTS RMS
1 41
(YOUR CHOICE r n
WHETHER CIV AT > 0.42 DC RATED VOLTAGE VOLTS RMS
10,50 OR 100 pC)
(4) THIS ENSURES THAT DURING SERVICE THERE WILL BE NO PARTIAL DISCHARGES.
Figure 5.5. AC PD testing.
Chapter 5: Partial Discharge or Corona Measurement
5-3
High Voltage Power Supply Design Guide for Space
B. Partial Discharge under DC Applied Voltage
Goddard Space Flight Center is involved in high voltage supplies and instruments, all of which are DC devices. For
testing of integrity of potting jobs or other insulation in and around DC image tubes or circuit components the ap-
plication of AC voltage is often not allowed.
The Theory shows that during quiescent DC application of voltage, the partition of it is according to the resistances
of the cavity and the intact dielectric in series with it (Fig. 5.2 and 5.6). Because the resistance of the high resistivity
dielectric is much greater than the resistance of the cavity, then on quiescent DC the cavity does not get much of the
voltage across it. Moreover, after the first discharge, recharging is by slow conduction through the highly resistive
dielectric. To quote from F. Krueger’s book, Discharge Detection in High Voltage Equipment: DC Voltage, “When
DC voltage is applied, discharges occur during the rise of voltage, as in the case of AC voltage. After the voltage has
become constant, discharges occur only infrequently” (Figs. 5.6 and 5.7). This does not give enough data, on quiescent
DC, during a reasonable observation time.
The question, therefore, is: How to change the DC partial discharge test to get enough data from it? The answer is to
collect data on the up-ramp, then separately at the V R plateau and then on the down-ramp. Measure pulse numbers, n,
and their magnitude in picocoulombs, pC, not the CIV. For applied voltage versus time, see Fig. 5.8.
Also, one must realize that insulating materials have a “memory,” especially ferroelectric materials such as ceramic
capacitors of barium titanate. Before testing these capacitors with DC, therefore, they must be heated at 85°C for 12 h
with leads shorted, to depolarize them from previous applied DC voltage.
• During quiescent DC voltage application, voltage partition is approximately
according to Resistances of Cavity, R c , and intact Dielectric in Series with it,
Rb (see Fig. 5.2).
• Rb of high resistivity Dielectric > R c of cavity.
• Therefore, on quiescent DC, the cavity does not get much of the applied voltage.
• After the first discharge pulse, recharging of the cavity is by slow conduction through
the Dielectric.
• Approximate discharge frequency / P.D. : according to Densley, Refs. [1], [11], [12],
and Rogers & Skipper, Ref. [23].
DC: / P.D. = 1_o E = 1.13 x 10 11 a E
®o Ej Ej
a = Conductivity of Dielectric
E = Stress in Dielectric
Ej = Discharge inception stress in Cavity.
(Stress = Electric Field)
e 0 = Permitivity of Empty Space
Figure 5.6. DC partial discharge (continued on Fig. 5.7, next page).
5-4
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
If E= Ej and o = 1 [Conathane]
4x10 13 Dm
Max. f = 1.13 xIO 11 1/s
P.D.s 4x10 13
= 3x10 _3 /s = 3 pulses/1000 s
= 3 pulses/16 min
• THIS IS NOT ENOUGH DATA TO BASE ANY CONCLUSIONS ON IN A REASONABLE
OBSERVATION TIME.
• DC CIV IS NOT A GOOD QUANTITY TO MEASURE, IN THE QUIESCENT DC APPLICATION.
Figure 5.7. DC partial discharge (continued from Fig. 5.6, previous page).
Applied
Voltage
Figure 5.8. Voltage -Time profile for abbreviated DC RD. test, Refs. [12], [21]. (The 60 s plateau at 1/2
rated voltage is omitted.)
Chapter 5: Partial Discharge or Corona Measurement
5-5
High Voltage Power Supply Design Guide for Space
Acceptance/Rejection Criteria forX7R, X5R BaTi03 Disc Capacitors of 1000 pF Capacitance
Note: Before DC PD Testing, the capacitors must be heated at 85°C, for 12 h. Leads shorted, to depolarize. Do this
every time before a DC PD test.
• On ramp-up to rated voltage:
(1) No single pulse to be greater than 150 pC;
(2) Sum of n^i to be no greater than 3000 pC.
• On quiescent plateau of rated voltage:
(3) No single pulse to be greater than 25 pC;
(4) In 100 s of observation, the Sum of nq, to be no greater than 150 pC, or “corona current” to be no more than
1.5 pC/s.
• On ramp- down to 0 kV:
(5) Very little, if any, relaxation corona is usually observed on ramp-down of these highly ferroelectric materials
like BaTi0 3 . Much relaxation corona has always been associated with macroscopic cracks. Therefore, no
more than 300 pC for sum of n,q,
• Also, no preferred peak distributions and no multiple corona bursts allowed.
• Naturally, different capacitances than 1000 pF have to be allowed different amounts of corona:
(6) On the sum of n q , , multiply by (C new /1000 pF) ratio.
(7) On the maximum charge content of single pulses, multiply by (C ncw /1000 pF) ' 2
(8) Any capacitor whose partial discharge is notably higher (~ half an order of magnitude higher) than
others from the same lot, should be rejected.
(9) NPO multilayer capacitors: No PDs on ramps or quiescent VR Plateau. Any other behavior is rejected.
(10) X7R Multilayer:
PDs on up-ramp.
Very little PDs on VR plateau.
Little PD on down-ramp.
Must be correlated/life tests, to get acceptance/rejection criteria.
Meantime, use similar criteria as for discs.
(1 1) Z5U Multilayer, and Z5U disc:
Much PDs on ramps.
Also PDs on VR plateau.
Do not use.
II. Examples of Partial Discharge Data
Of the many DC data runs taken over the years, the next page shows a typical one for an X7R BaTiO, multilayer ca-
pacitor in Table 5.1. This capacitor was rejected, having too high a sum of n^ on the ramp to rated voltage. Table 5.2
describes in words, that data one measures on a given set of capacitors when one chooses to do either a DC or an AC
test or both. Table 5.3 gives two columns on DC data and one column of AC data, taken on 24 single disc capacitors.
One can see that the lower half of the page of capacitors printed in Table 5.3 have much less picocoulombs for sum of
nqi on the ramp and 0 pC during the quiescent DC testing. The same lower half is also better on the AC test, in that
the corona inception voltage is higher. The only serial number that does not “lit in” is S/N 27, which looks like the
worst capacitor with lowest AC inception voltage of 1.4 kV rms, and yet it is not too bad on DC testing.
5-6
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 5.4 shows DC partial discharge data for a 3 kV DC, X7R, multilayer capacitor. Its enlarged Destructive Physical
Analysis (DPA) cross section photo of Figure 5.9. Table 5.5 is data for a 2 kV DC, X7R multilayer capacitor from a
production lot that had electrical failures. Figure 5.10 gives the enlarged photo of its cross section by DPA and reveals
several large pancake-shaped voids. This explains why Table 5.5 had so many numerous and energetic pulses as com-
pared to the capacitor in Table 5.4. But one can not DPA every capacitor. And so, the partial discharge test is a good
nondestructive method to reveal which capacitors will probably fail and which are risky to put into a circuit.
Table 5.1. Typical data for one multilayer capacitor; BaTi0 3 , X7R; 27,000 pF 1 kV DC;
Calibration: 4-500 pC: Amplifier: xlO
Time
Sec
kV
N
total
4-25
pC
n
25-100
pC
n
100-200
pC
n
200-300
pC
n
300-400
pC
n
400-500
pC
n
Max pC
Sum
n i4i
pC
12
0-.5
13
8
2
1
1
0
1
401
952
*12
.5-1.0
342
211
89
18
11
7
6+
4,819
179,437
60
1.0
6
2
3
1
104
283
30
1 . 0-0
23
14
6
2
1
246
802
* By changing the amplifier to xl before this ramp, the calibration becomes 40-5000 pC. This way, one “catches”
the big pulses.
Table 5.2. PD measurement: Correlation between DC and AC data.
DC DATA
AC DATA
(I)
One measures number of pulses,
the pC content of each and
computes sum of n^qj
One measures corona inception
voltages at a given pC level (100 pC
for X7R. 20 pC for NPO).
a)
b)
c)
On the ramp to V rated
At V rated for 100 s.
On ramp down to 0 V
CIV volts
Sum of njqj and max pC pulse- on
ramp to VR and at VR
(II.) Effect: In a set of capacitors, from
best to worst:
Sum of njqj goes up by factor of 6
or 7 on ramp, from 1,330 pC to
9,680 pC, for example.
CIV volts only decreases by 30%, that
is from 2.2 kV to 1.5 kV rms
Chapter 5: Partial Discharge or Corona Measurement
5-7
High Voltage Power Supply Design Guide for Space
Table 5.3. Correlation Between DC and AC Data; Single Disc D64 X5R 192M 5 kV, Maida; Post Burn-
In Corona Test.
DC DATA
AC DATA
S/N
0-5 kV in 20 s ramp;
Quiescent at 5 kV 100 s
AC corona
Sum of njqj on ramp
after wait 60 s
inception
(in pC)
(in pC)
voltage at 100 pC
(in kV rms)
1
8020 pC
369 pC
1.5, 1.65 kV rms
2
7525
69
1.6, 1.65
26
8980
0
1.7, 1.9
28
9680
21
1.8, 1.8
39
7520
12
1.6, 1.6
41
7580
0
1.8, 1.8
44
6050
11
1.7, 1.8
70
7820
0
1.65, 1.7
72
6280
28
1.8, 1.8
90
6810
74
1.6, 1.6
86
5480
0
1.7
55
5800
0
1.8
27
5730
0
1.4, 1.5
11
1960
0
1.9, 1.9
17
1828
0
1.9, 1.9
18
1330
0
2.1, 2.0
51
1480
0
2.0, 2.1
53
1750
0
2.1, 2.1
60
1650
0
73
1290
0
1.9, 2.1
83
1970
0
2.0
43
2220
0
2.1
49
2280
0
1.8, 1.8
84
2590
0
1.9
76
2900
22
2.0
15
3930
0
2.1, 2.1
5-8
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 5.4. One multilayer capacitor DC PD test; new Novacaps for Mr. Birsa; S/N 4.1 2020 B 332 M
302S, X7R, 3 kV DC.
Time kV
N 4-25 25-100 100-200 200-300 300-400
400-500
Max
Sum
Seconds
Total n* H n II n || n || n
n
pC
n i4i
pC
10 0-1.5
0
60 1.5
0
10 1.0-3
13 9 4
89.0
295
60 3
0
10(+40) 3-0
0
Reverse Polarity
0-1.5
1
22.0
22.0
1.5
0
1.5-3
5 3 1 1
143.1
226.7
3
0
3-0
0
* Where “n” is the number of pulses within the noted range. This was done at Calibration: 4-500 pC; Amplifier: xlO
Figure 5.9. Cross section of capacitor S/N 4.1, enlarged.
Chapter 5: Partial Discharge or Corona Measurement
5-9
High Voltage Power Supply Design Guide for Space
Table 5.5. Another multilayer capacitor DC PD test from the lot that had failures, for Mr. Birsa;
S/N 3.0, 2020B 103M 202S, X7R, 2 kV DC.
Time
Sec
kV
N 4-25
total pC
n
25-100
pC
n
100-200
pC
n
200-300
pC
n
300-400
pC
n
400-500
pC
n
Max
pC
Sum
n iMi
pC
10
0-1
484+ 319
57
40
24
25
23+
504+
37,013+
60
1
0
10
1-2
1597+ 896
603
74
22
7
6
4,844+
598,860+
60
2
5 2
3
75
205
10
2-0
329+ 138
93
54
22
14
14
511+
31,132+
(+40)
NOTE: This was done at Calibration: 4-500 pC; Amplifier: xlO
DO NOT USE THE ABOVE CAPACITORS
Figure 5.10. Cross section of capacitor S/N 3.0, enlarged. (Note pancake voids.)
5-10
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
REFERENCES
1. Densley, J., Engineering Dielectrics, Vol. I, STP 669, ASTM 1979, p. 409.
2. Dakin, D.W., Proc. 8 th Electrical NEMA-IEEE Insulation Conf, Los Angeles, California, Dec. 1968.
3. Bickford, K.J., and W.J. Sarjeant, 1981 Conf. Electr. Insul. Dielectr. Phenom., IEEE, CH1668-3, 1981,
p. 177.
4. Melville, D.R.G., B. Salvage, and N.R. Steinberg, “Discharge detection and measurement under direct Volt-
age conditions: Significance of discharge magnitude; Proc. IEEE 112, 1965, pg. 1815.
5. Densley, R.J., and T.S. Sudershan, “Partial discharge characteristics of solid insulation containing spherical
cavities of small diameters. NRC Conf. Electr. Insul. Dielectr. Phenomena. Oct. 1976, National Academy
of Sciences, 1977.
6. ASTM, Detection and Measurement of Discharge (Corona) Pulses in Evaluation of Insulation Systems. D
1868-1881, 1981.
7. F. Krueger, Discharge Detection in High Voltage Equipment. American Elsevier Publ., 1968.
8. Parker, R.D., “Corona Testing of High Voltage Airborne Magnetics,” Proc. 1975 Power Electronics Special-
ists Conf, IEEE 1975, pg. 43.
9. Corona Detection in Insulation Systems. Biddle Technical School Text, Blue Bell, Pennsylvania, Feb. 1970
10. IEEE, “IEEE Recommended Practices for the Detection and Measurement of Partial Discharges During Dielectric
Tests,” IEEE Std. 454-1973, 1973.
11. Hai, F., and K.W. Paschen, “Development of a Partial Discharge Detection System for Traveling Wave Tube
Testing,” Aerospace Corp. for Air Force Systems Command, Rep?. #SAMSO-TR-79-40, Sept. 28, 1979.
12. Bever, R.S., and J.L. Westrom, IEEE Trans. AES, Vol. AES-018, No. 1, Jan. 1982, pg. 82.
13. Bever, R.S., B. Seidenberg, and J.L. Westrom, “High Voltage Testing of Witness Samples for Faint Object
Camera of the Space Telescope Project,” X440-82-8, NASA Goddard Space Flight Center, Greenbelt, Mary-
land, April 1982.
14. Dunbar, W.G., and P.A. Tjeller, “Manufacturing Technology for Airborne High Voltage Power Supplies,”
Vol. I, Boeing Aerospace Co., Tech. Rept. AFML-TR-79-4018, Vol. I, AFML, U.S. Airforce, Feb. 1979.
15. Burnham, J., “Recent Advances in Interpretation of Corona Test Results,” Proc. IECEC Conf, Anaheim,
California, Aug. 1982.
16. Dakin, T.W., “Partial Discharges with D.C. and Transient High Voltages,” Proc. Nat. Aerospace Electronics
Conf, Dayton, Ohio, May 1978.
17. Jackson, R. J., A. Wilson, and D.B. Giesner, “P.D. in Power Cable Joints,” IEEE Proc., 127C, 1980, 420-429.
18. Parker, R.D., R.V. Delong, and J.A. Zdik, “Accurate Corona Detector Calibrator,” IEEE Trans. Electr. Insul., EI-15,
1980, 451-454.
19. Harrold, R.T., and T.W. Dakin, “The Relationship Between the Procarbons and the Microvolt for Corona
Measurements on HV Transformers and Other Apparatus,” IEEE Trans. PAS-92, 1973, 187-198.
20. Zaengl, W.S., P. Osvath, and H.J. Weber, “Correlation Between the Bandwidth of P.D. Detectors and Its
Inherent Integration Errors,” Proc. 1986 Interntl. Symp. Electr. Insul., Washington, DC, June 1986.
21. Bever, R.S., “Ramp Technique for DC Partial Discharge Testing,” IEEE Trans. Electr. Insul., EI-20, No 1,
1985, 38-46.
22. Hai, F. and K.W. Paschen, “Partial Discharge Tests of Multi Winding High Voltages Transformers for Space
TWTA’s,” Aerospace Corp. Rept. SD-TR-85-65, for Airforce Systems Command, Sept. 1985.
23. Rogers, E.C., and D.J. Skipper, “Gaseous Discharge Phenomena in High-Voltage d.c. Cable Dielectrics,”
Proc. Inst. Electri. Eng., 107, 1960.
Chapter 5: Partial Discharge or Corona Measurement
5-11
High Voltage Power Supply Design Guide for Space
APPENDICES TO CHAPTER 5
Appendix I to Chapter 5: Biddle* Co. Partial Discharge Detection Equipment.
This is included as a help to new technicians at Goddard Space Flight Center.
*The James G. Biddle Co. has gone out of business. Repair on equipment can still be obtained.
Appendix II to Chapter 5: Detailed Directions for Partial Discharge Measurement, for Capacitors.
This is included as a help for new technicians at GSFC.
Appendix III to Chapter 5: Directions for GLAST Project Flex Cable Corona Test
This is an example of corona testing of different items than capacitors.
Appendix IV to Chapter 5: MIL-PRF-49467B; 2 May, 2001, Directions for AC Corona Testing of Multilayer
Ceramic Capacitors.
This illustrates the approval of this type of testing.
Appendix V to Chapter 5: AC Corona Test of Transformers at 50-100 kHz
By, J.D. McCormick and M. Mogavero, Westinghouse Corp., Baltimore, Maryland, S-311320-CAPS/IS50
transformer, Sept. 1994. This illustrates the corona testing of transformers at closer-to-actual operating
frequency than with the usual 60 Hz commercially available equipment.
5-12
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Appendix I to Chapter 5
BIDDLE* CO. PARTIAL DISCHARGE DETECTION EQUIPMENT
*The James G. Biddle Co. has gone out of business. Repair service can still be obtained.
I. The Biddle equipment can be in one of the three possible configurations:
(1) DC only
(2) AC only, at 60 Hz.
(3) AC and DC superposed
It is often in either (1) or (3). If in (3), (the superposed mode), the voltage limits are: 12 kV rms maximum as read on the
RED display and 38 kV DC as read on the RED display on the AC control cabinet.
Also, if in (3), then the DC ammeter will register some current when DC voltage is applied.
The diagram on next page illustrates the connections in the test cabinets, for modes (3) and (2) above. You must open
all the test cabinet doors to inspect and make changes if desired. See Figures 5.A.I.1 and 5.A.I.2.
II. Read the Biddle books (3 maroon-colored direction books).
III. It is imperative that no ordinary cloths or paper towels or chamois cloth be used inside the test cabinets. Only
lint-free cloth obtained from GSFC Materials Branch is permitted (small hard cotton squares). Damage to
the Biddle system is “murder” to repair because of the size of the equipment. A Moving Van is required to
return it for repair.
IV. The polarity of the high voltage output can only be changed by opening the DC HV supply and reversing
some jumpers under the oil using clean plastic pliers. Only experienced people are allowed to do this. No
dust or flakes of skin must be allowed to fall into the oil during this operation.
Chapter 5: Partial Discharge or Corona Measurement
5-13
High Voltage Power Supply Design Guide for Space
Figure 5.A.I.I. Bus arrangement for AC and DC superposed voltages. Maximum is AC 12 kV rms 60 Hz,
DC 38 kV DC, Ref. [9].
5-14
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Appendix II to Chapter 5
SOME DETAILED DIRECTIONS FOR PARTIAL
DISCHARGE TESTING, BOTH DC AND AC
General Comments
Before you start, you must know:
1. The condition of your equipment: Is it connected for the DC mode or for the superposed mode which is used for
AC PD measurements? What is the polarity of the high voltage terminal with respect to ground? All the settings
of the amplifiers, calibrator, control knobs need to be inspected and be set to your needs. You cannot assume that
any of the settings are still the same as when you last used the equipment.
2. The properties of the materials or assembly you are testing: all insulating materials have a memory, that is, if high
voltage has been applied previously in a certain polarity the material remains polarized in that polarity, even if
the leads to it have been shorted at room temperature. This is especially true of ferroelectric materials, such as the
ceramic capacitors (except the NPO formulation). Therefore, before PD testing, these materials need to be heated
with leads shorted, to obtain consistent results from one test to the next. Despite this depolarization procedure,
there is still some remnant memory and the sample should always be connected in the same polarity, so as not to
introduce an unwanted variable of polarity reversal.
The conductivity and the dielectric constant of the material have a vast influence on the partial discharge to be
expected. The greater the conductivity, the more partial discharges are to be expected on DC testing for a given
defect site. Also the higher the dielectric constant, for a given externally applied voltage and a given defect, PDs
will incept at a lower applied voltage, both on AC and on the ramps during DC testing.
This means that you cannot blindly apply the same acceptance/rejection criteria to everything you test. Therein
lies the great difficulty of interpreting the PD data.
3. The particulars of the equipment are contained in the manuals supplied by the manufacturers of it, and kept in
the file cabinet in the high voltage laboratory. There is a DC System, an AC System, a High Voltage DC power
supply, and an AC supply, a Multichannel Analyzer (MA) and a Computer System. Because the last two items are
probably different for each laboratory, then detailed directions must be supplied for the technicians there.
I. DC PARTIAL DISCHARGE MEASUREMENT
*The James G. Biddle Co. has gone out of business, however, the equipment is still fully functional.
A. De-polarization De-aging
Ceramic capacitors, being mostly BaTi0 3 with additives, are highly ferroelectric. Therefore, they always need to be
de-polarized or de-aged before DC partial discharge measurements (DC PD) are taken. Mount the capacitors in a
shorting jig. Be careful not to mechanically stress the leads where they exit from the capacitor body. Place in a clean
oven. Raise the temperature in about 15 min to 85°C and hold at 85°C for a total of 12 h. Cool back to room tempera-
ture in about 15 min. The 12 h of heating need not be continuous, e.g., it can be 7 h and 5 h.
The capacitors will still have a small amount of polarization after this treatment. It, therefore, is important to be con-
sistent in polarity. In the DC PD measurements, be sure to connect the leads in the same polarity as the last previous
Chapter 5: Partial Discharge or Corona Measurement
5-15
High Voltage Power Supply Design Guide for Space
high voltage connections, e.g., as in the DWV High Voltage power supply or in the Voltage Conditioning High Voltage
power supply. If connections are made in reverse, make a note of it.
B. Calibration
This must be done with great care every time a new batch of capacitances is DC PD measured. This is because the
sensitivity of the equipment depends on the test sample capacitance. Every time the test sample capacitance is changed,
one must recalibrate.
In the Biddle Test Cabinet
Be sure you know the polarity of the high voltage terminal with respect to ground. Connect the test sample capaci-
tor in the same polarity as in the last previous high voltage exposure or know if reversed. Now, also connect the low
voltage calibration capacitor, if no high voltage calibration capacitor is in the test cabinet. Connect its B and C cable to “Cal.
Signal out”. (Be sure after the calibration is accomplished, as below, to remove the low voltage calibration capacitor.)
If, however, a high voltage calibration capacitor is in the system, omit the last two steps.
On the Biddle Control Console
Have the amplifier on the xlO position and switch on the Direct Calibration. Two separate calibration signal peaks should
now be visible on the oscilloscope (each peak also has a small secondary peak at its foot). Go by the higher peak.
Now dial up the desired amplitude of the calibration signal peaks. Be sure to do this or check on it because it may
have been changed by someone. Suggested values are given below.
Suggested Calibration Signal Amplitudes
NPO X7R <10,000 pF X7R >10,000 pF
200 pC 500 pC 1000 pC
The values above are merely suggestions for the very first time that a given batch of test capacitors is being DC PD
tested. The aim is to catch the most energetic pulses while at the same time not cutting off the lowest ones. Once the
first capacitors of a given batch have been satisfactorily measured, then on later measurements on the same group,
the same calibration signal amplitude must be used. Remember also that once the high voltage has been raised on a
given capacitor, you cannot go up on it again in the same polarity without de-polarizing. You can reverse polarity and
go up again, but the PD pulses will be approximately double (hysteresis in ferroelectric materials).
Now, change the Amplifier Amplitude Vernier until the highest of the two peaks is 6.5 cm high. This corresponds to
the full scale on the x-axis of the MA, which is used as the quantitative data read-out instrument for DC PD measure-
ments. You will have to fine-tune the 6.5 cm after turn-on procedures on the MA turn-on procedures, so do not yet
remove the low voltage calibration capacitor.
On the Multichannel Analyzer
Detailed directions must be obtained in the particular laboratory. The MA has two markers that move on the horizontal
axis, and the right-hand marker in the extreme position on channel 1024 should correspond to the vertical calibration
signal amplitude chosen for the Biddle vertical pulse height on the Biddle oscilloscope. This calibrates the horizontal
position of the markers in picocoulombs, and one can thus read off the number of pulses, n, between certain pico-
coulomb values.
5-16
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
When calibration is complete, remove the low voltage Calibration Capacitor from the Biddle Test Cabinet and tighten
the corona ball on the HV terminal.
C. DC Partial Discharge Measurement (DC Ramp Test)
Be sure the low voltage calibration apacitor is removed.
Be sure the test sample ends of leads are caught into and buried inside of the alligator clip jaws of the high voltage
connections to the Biddle terminals. No cut ends of terminals must stick out. Be sure the entire test sample and all
naked metal of the connecting cables and fixturings are immersed in Fluorinert liquid FC-43 if the voltage is going
to be 2kV or higher. Actually, this should also be done at lower voltages. (The sample can also be placed in high
vacuum.)
1. Set up regions of interest between markers on the MA according to your instrument’s directions.
2. Be sure on the Biddle cabinet the voltmeter switch is in the lowest position. The variac must be turned full
down or counterclockwise. Turn on the high voltage. The red light should come on.
3. Acquire pulses for 30 s at 0 V. There should be no pulses unless there is undesirable noise on the electric
line. (In the past, there has been noise below channel 13 or so, but you can arrange not to count it by having
the left maker on channel 14.)
4. Acquire pulses while slowly raising voltage to ‘A V-rated in 10 s and acquiring for 2 s more. Record the
total number of pulses; also record the pulses within the regions of interest and the maximum pulse value in
picocoulombs.
5. If in step 4 there were pulses above A full scale, then chances are that during the next ramp to full-rated voltage
for X7R capacitors, there will be pulses in the several thousand picocoulomb range.
To catch these high pulses, turn the amplifier gain to xl. This now will, of course, make the MA x-axis go from
45-5000 pC, instead of 4.5-500 pC. Because you changed the MA, now you must multiply the picocoulomb values
by 10 before recording. Acquire pulses while raising the voltage in 10 s to V rated. Wait 2 s more. Record all data as
in Step 4.
6. Return the amplifier gain switch to xlO. Be sure to do this. The previous data recording should take up to 60
s of waiting period between taking the hands off the voltage variac and doing data acquisition for the truly
quiescent V rated plateau. So now, at V rated, acquire pulses, and let run until the preset time of 60 s stops
the acquisition. Record all data as before.
7. Acquire pulses, data while running the voltage variac down as fast as desired. Then, wait out the relaxation
pulses for about 30 s more. Stop. Record all data as before.
8. Turn off the High Voltage. Discharge the HV terminal in the test cabinet with grounding wand. Remove the
test sample and replace it with another sample. Close door. Repeat entire procedure III from step 2 onward
until a different value of capacitance comes up at which time the calibration procedure has to be repeated,
as in section II.
After practice, the entire measurement should take no more than 5 min per sample.
Chapter 5: Partial Discharge or Corona Measurement
5-17
High Voltage Power Supply Design Guide for Space
Table 5.A.II.1. Typical data DC PD for a single capacitor; multilayer BaTi0 3 , X7R; 27,000 pF,
1 kV DC
Time
Sec
Amplifier
Gain
kV
n
4-25
n
25-100
n
100-200
n
200-300
n
300-500
n
Max pC
Xmi
pC
12
X10
0— >.5
13
8
2
1
1
1
401
952
12
XI
,5-M.O
342
211
89
18
11
12
4,819
179,437
60
X10
1.0
6
2
3
1
104
283
30
X10
O
t
o
23
14
6
2
1
246
802
Table 5.A.II.2. Abbreviated version typical DC/PD data for two capacitors. Multilayer BaTi0 3 , 820 pF,
3 kV DC
MFR: WRT SET #2 S/Ns: 2209-2210; Cap: 820 pF; Vrated: 3 kV; Dielectric: X7R; Calibration:
4-500 pC; Amplifier: X10
Results: Accept S/N 2209 because Xnqi -3000 pC.
Reject S/N 2210 because £ n iCji too high on ramp-up.
D. DC: Approximate Acceptance/Rejection Criteria for 1000 pF, X7R, X5R BaTi0 3 Disc
Capacitors
• On ramp-up to rated voltage:
(1) No single pulse to be greater than 150 pC;
(2) Sum of n,q, to be no greater than 3000 pC;
• On quiescent plateau of rated voltage:
(3) No single pulse to be greater than 25 pC;
(4) In 100 s of observation; the Sum of n,q, to be no greater than 150 pC, or “corona current” to be no greater than
1.5 pC/s.
• On ramp-down to zero:
Very little, if any, corona is usually observed on ramp-down in these highly ferroelectric materials.
Much relaxation corona has always been associated with a crack in the BaTi0 3 . No more than Sum n q.
= 300 pC.
• Also, no preferred peak distributions and no multiple corona bursts allowed.
• Naturally, different capacitances than 1000 pF have to be allowed different amounts of corona, as follows:
(5) On the Sum of n,q„ multiply by (C new /1000 pF) ratio.
(6) On the maximum charge content of single pulses, multiply by (C ncw /1000 pF)' /2 .
• Finally, any capacitor whose partial discharge is notable higher (half an order of magnitude higher) than others from
the same lot should be rejected.
5-18
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
II. AC PARTIAL DISCHARGE MEASUREMENT
A. De-age
Same as in DC; it is not absolutely necessary, but would be advisable for the sake of consistency.
B. Calibration
Same as in DC. Be sure it is done AFTER the extra bus-bar has been connected from the A.C. cabinet
to the High Voltage terminal in the D.C. cabinet. The Multichannel Analyzer is not absolutely necessary
for the A.C. measurement, but is nice to turn on to determine when 100 pC pulses appear with greater
accuracy than by just watching the Biddle oscilloscope. For A.C. corona measurements, calibrate with
200.0 pc calibration signal injection for all capacitors. In this way, the 100 pc pulses will be about 3 cm
high on the Biddle oscilloscope and will be Zz of the way along the MA screen, and easily visible. For
NPO capacitors, 25 pC inception voltage should be looked for and recorded.
C. High Voltage AC PD Measurement
Turn lights off in the room after sample is connected. Turn on the AC High Voltage. Slowly raise the AC
High Voltage Variac while also watching the Biddle oscilloscope and also the red test sample voltage
readings on the digital voltmeter on the Biddle AC control cabinet. When pulses appear on the Biddle
oscilloscope that correspond to 100 pC, then stop raising the voltage. (If 200 pC = 6.5 cm, then 3.25 cm
= 100 pC.) Alternatively, have the MA in the Acquire mode with infinite running time and watch for
pulses appearing beyond the 100 pC marker. Stop raising the voltage, and read the red digital voltmeter.
Turn voltage variac down immediately. The voltage that was recorded is the Corona Inception Voltage
or CIV. Repeat at least once. The CIV is defined as the voltage at which pulses higher than 100 pC (or
25 pC for NPOs) keep coming in for 5 s. Average the two voltage readings. Do not leave the voltage on
where 100 pC pulses persist for longer than 5 s. The CIV is, of course, an AC rms voltage.
D. Voltage Limit
This is important for NPO capacitors. For these capacitors, it is possible that the CIV is quite high. Also,
the X7R capacitors may be better than previous experience indicated. AC applied voltage on all of these
capacitors may be damaging because of heating effects. Therefore, the following maximum voltage
should not be exceeded:
Maximum AC rms volts= 70% of DC V rated = .49 DC V rated allowed to be applied
1.43
For a 5 kV rated DC Capacitor this equals 2.5 kV rms AC
for 2 kV = 1.0 kV rms
for 1 kV = 0.5 kV rms
D. Acceptance/Rejection Criteria: Approximate
For X7R, X5R, Z5U: Reject if, at 100 pC levels, CIV <50% DC V r in volts rms
For NPO, SrTi0 3 : Reject if, at 25 pC level, CIV <50% DC V r in volts rms
Chapter 5: Partial Discharge or Corona Measurement
5-19
High Voltage Power Supply Design Guide for Space
Appendix III for Chapter 5
DIRECTIONS FOR CORONA TEST OF FLEX CABLE BOARD
Flex Cable Board Corona Test
Corona testing will be performed to detect defects within the flex cable configuration. One series of tests will be
conducted to examine the cabling between boards. The second series will test for defects between layers within the
boards.
GLAST Photomultiplier Flex Board PWB 2054578 Rev - 02/9/04
Primary Component Side Flex Layer 2
Layer 1 of 4 Layer 2 of 4
Flex Layer 3 Secondary Component Side
Layer 3 of 4 Layer 4 of 4
Figure 5.A.III.1
The Resistor Network flex-cable version is shown above: 14-in Ultem spacers will be used to hold the boards in flight
configuration. As this test is strictly making an evaluation of the board itself, no components will be mounted. The
Code 563 Biddle Corona Test System in Building 20 will be used for corona detection. Testing will cover the charge
range from 10 pC to 500 pC. The sample will either be tested in Fluorinert FC43 or in vacuum at pressure less than
10 x 10E-6 torr.
Clean gloves will be used when handling the boards for all operations.
#24 AWG bus wire will be installed in the following tube pin locations: Pins 5, 6, 7, 8, and 13.
A test lead will be installed on C5+.
5-20
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 5.A.III.1. Apply voltage according to the table below and record data in the test sheet.
Testing
+HV
Ground
Between cable layers
Pin 8
Pin 13
Within cable layers
Pin 7
Pin 5
Board to cable
C5+
Pin 6
Board top to
board bottom
C5+
Pin 8
GLAST Photomultiplier Flex Board
PWB 2054578 Rev-02/9/04
% /•
Primary Component Side
Layer 1 of 4
Figure 5.A.III.2
Chapter 5: Partial Discharge or Corona Measurement
5-21
High Voltage Power Supply Design Guide for Space
Table 5.A.III.2. Flex-cable resistor network corona test. Apply voltage according to the table and record
data in the test sheet.
Test
Points
Corona Counts
0-2 kV 2 kV
2 kV 4 kV 4 kV 6 kV 6 kV 8 kV 8 kV 10 kV
to 4 kV to 6 kV to 8 kV to 1 0 kV
Pin 8 to
Pin 13
Pin 7 to
Pin 5
C5+ to
Pin 6
C5+ to
Pin 8
5-22
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Appendix IV for Chapter 5
AC Corona Test for Capacitors, Fixed, Ceramic, Multilayer, High Voltage (General Purpose) EXCERPTED FROM
MIL-PRF-49467B; 2 May 2001
MIL-PRF-49467B
AC PARTIAL DISCHARGE (CORONA) TEST
B.l SCOPE
B.1.1 Scope. This appendix details the detection and measurement of partial discharge (corona) I in 3.10 and 4.8.6. This
appendix is a mandatory part of the specification. The information contained herein is intended for compliance.
B.2 APPLICABLE DOCUMENTS
This section is not applicable to this appendix.
B.3 REQUIREMENTS AND DEFINITIONS
B.3.1 Supply Voltage. The supply voltage for AC partial discharge tests shall be variable AC voltage at a frequency of
60 Hz ±5 percent and shall be measured in AC volts rms with a tolerance of ± 5 percent of the AC test voltage.
B.3. 2 Sensitivity. The partial discharge detection system’s sensitivity depends on the capacitance of the test specimen.
The test specimen shall be connected during system calibration and the sensitivity requirements shall be:
a. For capacitances of less than or equal to 0.005 pF, system sensitivity shall be able to detect 5 pC or less;
b. For capacitances above 0.005 pF, to and including 0.1 pF, system sensitivity shall be able to detect 15 pC or
less;
c. For capacitances above 0.1 pF to and including 0.47 pF, the sensitivity shall be 50 pC or less.
B.3. 3 Corona inception voltage (CIV). CIV at a given pC level shall be defined as the voltage at which continuous
partial discharges can be recorded at that pC level. This is above the minimum sensitivity, as the applied voltage is
increased at a constant rate.
B.4 TEST CONDITIONS AND PROCEDURE
B.4.1 Connection. The capacitor under test shall be connected between the high voltage terminal and ground of the
detection system with insulated, corona-free cables. The capacitor, its leads, and bare metal connecting clips shall be
immersed in FC-40 or FC-43 dielectric fluid (fluorinert) or equivalent.
B.4. 2 System calibration. The system shall then be calibrated in accordance with the requirements of B. 3.2.
BAB Application of voltage. The applied voltage shall then be increased at a constant rate of approximately 0.2 kV
rms/second. The maximum test voltage shall not be more than 42 percent of DC rated voltage.
BAA Measurement. The maximum voltage specified in B.4.3 shall be maintained from 1 second to 5 seconds. If the
maximum corona pulse exceeds 100 pC amplitude in this time period, the component shall be considered to be a
failure. The voltage shall then be decreased to 0 volts.
Chapter 5: Partial Discharge or Corona Measurement
5-23
High Voltage Power Supply Design Guide for Space
Appendix V for Chapter 5
AC CORONA TEST OF TRANSFORMERS AT 50-100 KHZ FREQUENCY
By, J.D. McCormick and M. Mogavero
Westinghouse Corp., Baltimore, Maryland
S-311320-CAPS/IS50 transformer, Sept. 1994
CORONA TEST PLAN FOR N.A.S.A. TRANSFORMER (with actual changes in parentheses on schematic — see
page 2)
I. TEST
A) Demonstration was performed this date of the Corona Test described in Ref. 1 attached. Deviations taken
from the described procedure included:
1) The ITigh Pass filter was set for 200 kFIz instead of the 500 kFlz described to make sure low frequency
corona did not get filtered out, as corona frequencies can start as low as 300 kFlz.
2) C-l was a 22 pF, 4 kV capacitor instead of the calculated value of 80 pF and the resonant frequency for each
unit was still about 1/2 of the goal of 75 kHz. This was due to the internal capacitance of the transformer as
well as the Power Separation Filter in the test set that tended to allow the circuit to load at higher frequencies
making it impossible to drive it to the specified value of 2000 V p-p above about 40 kHz with the existing
400 W amplifier. (Additional work could have possibly increased this frequency but the ability to reach 75
kHz is in doubt.)
It must also be noted that the circuit sketch had an error that was pointed out by Art Ruitberg from NASA. The “Cal
Signal” was injected into the junction of C-l and the UUT rather than the top of the series resonant circuit (note cor-
rection).
B) Test results were (see attached photographs):
1) A 5 pC calibrated signal measured 15.2 mV on the oscilloscope.
2) The first unit demonstrated had 24.6 mV of corona. It must be noted that there were air bubbles in the
mold material around the secondary coil of this unit and poor adhesion of the material to the inside of
the Ultem case. For these reasons a third secondary coil was manufactured.
3) The second unit had 6.08 mV of corona.
4) The secondary coil from the first unit was removed and replaced with another that had just completed
manufacture with the new secondary coil the transformer had 6.40 mV of corona.
II. MANUFACTURE
In addition to the manufacture of a third secondary coil as described above, several other surprises and deviations
from the quote (ref. 2) must be noted.
A) Additional Machining Steps:
1) The Housing, Transformer Mold, drawing 1308772 identified a qty. of two .055 inch dia. holes to
allow for the .062 inch dia. high voltage leads to penetrate the case. The cases were returned to the
machine shop to have these holes enlarged to .062 in. after telecon with Ken at N.A.S.A. verified that
this problem had been recognized by them.
5-24
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
2) The secondary coils that were wound by Westinghouse personnel at N.A.S.A. and verified by them, had a
height slightly higher than the top of the Ultem case (.469 in. coil height vs. a case height of .460 in.) after
they were glued into the cases. Request from N.A.S.A. by telecon to have .030-040 in. of mold material
above the coil wire resulted in the need to gently push the secondary coil wires down as far as possible after
they were glued into the case (resulting in getting the wire just at the top of the case) and then building up
the case with a temporary dam so that the material extended above the top of the case. The internal spacing
between each core was measured and the mold material machined down to allow .002 in. clearance between
the secondary and the core. This resulted in the molded secondary coil for unit number 2 being .492 in. (.032
in. of material above the coil wire) and molded secondary number three being .494 in. (.034 in. of material
above the coil wire). It must be noted that this was a relatively expensive deviation as Westinghouse expected
to pour the cases to the top with no further treat nor machining. The Ultem case had to be remachined to its
original dimensions as it was almost impossible to remove the mold material from the outside of the case.
B) Materials
1) All materials listed in the fabrication specification as well as the solder terminals were to have been
supplied by N.A.S.A. (see ref. 2 first bullet and N.A.S.A. drawing 1308769). This was not always the
case. The 200 proof alcohol, the Ambion Insulstrip, the Stycast 3050 mold material, the Epon/Versamide
adhesive and the solder terminals were all acquired through Westinghouse.
2) The Stycast 3050 was a very difficult material with which to work. It tended to explode into the vacuum
chamber and adhere strenuously to anything it contacted including the mold release. .Although several meth-
ods were used including immediate insertion into an 80°C. oven and several hours of room temp, settling
prior to insertion, Westinghouse has not yet gotten a good handle on the proper method to use to make sure
that there are no voids in the material. Art Ruitberg felt that the 80°C. cure temperature may not have been
correct although that was received from Grace by telecon. The second unit molded had air bubbles around
the coil and poor adhesion to the case as described above. The third secondary had a hole in it that was filled
with deaired material and recured. After this second cure the Ultem case cracked. We need to either get a
process or spend some time to develop one prior to continued use of this material. It must be noted that in
telecon with Grace Chemical, they suggested a Stycast 2850 material as it has better high voltage dielectric
and thermal transfer characteristics.
C) Processes
Westinghouse expected to have N.A.S.A. processes and procedures already developed for the application of adhesives
and molding (ref. 2 fifth bullet). In reality it was necessary to acquire this information from the manufacturers and
generate our own information for manufacturing personnel.
D) Assembly
The assembly generally went well. Other than the hole for the high voltage wire, the only other problem was that the
high voltage wire was difficult to ball solder and three of the four operators who tried (all with current Westinghouse
high voltage solder certification) had trouble getting a good solder ball. It was decided to attach the primary wires to
the terminals during assembly due to their size (AWG #34) and the fact that they would be needed to perform voltage
ratio test. All solder attachments to the high voltage wire were made using a low corona solder ball per Westinghouse
process. These were inside the secondary coil case. The primary coil solder attachments to the terminals were made
with best commercial practice.
Ill) SUMMARY
As with any development task, this one had deviations from what was expected and surprises in results. Excellent
responses from N.A.S.A. and material suppliers were a great help in getting a good handle on these deviations and
surprises. Conversation with Art Ruitberg from N.A.S.A. identified a stress relief that should have been applied to the
Ultem case after machining and prior to use. Application of this stress relief may have eliminated the cracked case.
Chapter 5: Partial Discharge or Corona Measurement
5-25
High Voltage Power Supply Design Guide for Space
Future use of this material will have stress relief performed. A hold on the charge number for two months was a slight
perturbation but all material and information were held ready so that little was lost once the spending was approved.
The corona test demonstration went well after some initial start-up problems although, as described above, the test
frequency was lower than anticipated. NASA has accepted the units with a few comments concerning poor machining
quality that will be addressed with the machine shop. (Some of this may have been the result of machining the mold
material from the outside of the case.) The question of solder certifications must be addressed prior to manufacture
of flight units. NASA, requested a ship-in-place status as they have additional work they would like Westinghouse
to perform. Overall this was a good learning experience for Magnetic Devices and we look forward to a continued
relationship in the space arena.
Sketch of Resonant Circuit Method of Corona Test
Select value of Cl such that it makes a series resonant circuit with the UUT at the desired test frequency.
At / : coL = 1 where ro=2ir/ and L= 74 mHz
° roC
C = _1_ - 80 pF (value actually used in circuit was 22 pF)
co 2 L
Figure 5.A.V.I. Resonant circuit method of corona test.
5-26
Chapter 5: Partial Discharge or Corona Measurement
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
First unit with secondary coil
that was replaced.
Corona = 24.6 mV
Figure 5.A.V.2
Second unit.
Corona = 6.08 mV
Figure 5.A.V.3
New secondary on first unit.
Corona = 6.40 mV
Figure 5.A.V.4
Chapter 5: Partial Discharge or Corona Measurement
5-27
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 6. Electrostatic Field Analysis
I. Physics and Mathematics Background:
One may analytically calculate the electric fields due to high voltage differences between ground and high voltage
terminals. This can be for parts of instruments or within high voltage power supplies. If the geometry of the terminals
or electrodes is simple, such as cylinders, spheres or parallel planes, then formulas exist for algebraic calculations.
Even for more complicated configurations there are expressions such as the Table 2.2 of Chapter 2, which gives the
Maximum field strength at the surface of the electrodes and perpendicular to them in direction. Electric fields are the
gradients of the equipotential lines and are perpendicular to them, everywhere.
If, however, the shape of the electrodes is complex and accuracy is required, then a computer analysis of the entire
space between and around the electrodes is indicated. The geometry of the electrodes can be 2-dimensional, of
infinite extent in the third, or it can be a 3 -dimensional solid of revolution, generated by revolving an x-y cross-sec-
tion about the y-axis. For computer analysis of electric fields, the so-called finite element method is best suited. This
is implemented in the well known, proprietary MSC/NASTRAN computer program (McNeil Schwendler/NASA
Structural Analysis). It was developed under the direction of Dr. William Case during his years at NASA Goddard
Space Flight Center, and it contains a Fleat-Flow portion. It is the NASTRAN Thermal Analyzer (NTA) portion of
the program which is of interest here. This is because the steady state differential equations governing the physics of
thermal gradients and of voltage gradients, are entirely analogous in mathematical form. Also, the ratio of potential
gradients perpendicular to the boundary in adjacent insulating materials is inversely proportional to the ratio of di-
electric constants. Similarly, the ratio of temperature gradients in adjacent materials is inversely proportional to the
ratio of thermal conductivities. So dielectric constant or rather, permittivity of an insulating material is analogous to
thermal conductivity (Table 6.1(a)).
If the dielectric material also has appreciable electric conductivity, then the steady state electric field distribution in
adjacent different materials is determined by the fact that the electric conductivity is analogous to thermal conductiv-
ity; then use the third part of Table 6.1(a).
Heat: For the steady-state solution of iso-thermal lines, LaPlace equation is
(
K
dx 2 + dy 2 + dz 2
\
T+Q = 0,
where
T = temperature in °K
Q = heat/second or joules/second generated per m 3 = W/m 3
K = thermal conductivity in J/m 2 = W
s °C/m °C m
1_* Q = °C/m 2
K
Electrostatics: For the steady-state solution of equipotential lines, LaPlace’s equation is, if electric conductivity is
zero:
e
y dx 2 + dy 2 + dz 2
T
/
R+p =0,
Chapter 6: Electrostatic Field Analysis by Computer
6-1
High Voltage Power Supply Design Guide for Space
where
V= voltage in volts;
e = K e e 0 , where K e = dielectric constant, e 0 = permittivity of empty space, and e = permittivity of the di-
electric in farads per meter (F/m), which is coulombs per volt meter; and
p= space charge density in coulombs per cubic meter (coulombs/m 3 ), if “doping” is used.
Thus, p/e = V/m 2
If answers for the field strength, E, are finally desired in volts per millimeter (mm) (millimeter are more the dimensions
of apparatus and instruments), then p has to be in coulombs per cubic millimeter and e 0 has to be changed from the
MKS value of 8.9 x 10' 12 F/m, to 8.9 x 10' 15 F/mm. The analogies are summarized in Tables 6.1a and 6.1b. Different
authors use different symbols for physical quantities.
TABLE 6.1a. Electrical/Thermal analogy for steady-state analysis.
Thermal
Electric
(if electric conductivity = 0)
T = Temperature, i.e., °K
K = Conductivity, i.e., W
mm °K
Q = Heat Source Density
i.e., W
mm 3
V = Voltage, i.e., volts
e = Permitivity, i.e., F/mm
= K e £ 0 = K e x 8.9 x 10" 15 F/mm
= Charge Source Density or Space
Charge, i.e., C
mm 3
K e = Dielectric Constant
KV 2 T + Q = o
e V 2 V + p = 0
VT = Thermal gradient
VV = -E, the Electric Field
Electrical (with conductivity not =0)
V=Voltage, i.e., volts
o=Electric conductivity (i.e., 1/Qm)
aV 2 V=-dp/dt
(Use SOL 59 when right-hand side is non-zero. Use HEAT SOL 24 for steady state current flow.)
6-2
Chapter 6: Electrostatic Field Analysis by Computer
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
TABLE 6.1b. Analogous thermal and electrical parameters, adapted from Ref. [10],
MSC/NASTRAN
Name
Thermal
Symbol
Electrostatic
Symbol
Current Flow
Symbol
Temperature
T
V
V
Conductivity
K
e
a
Temperature Gradient
VT
-E
-E
Heat Flux
KVT
D
J
Internal Heat
Generation
q_
P
dp/dt
Total Heat Flow
Q
Qe
I
II. EXAMPLE: From a report by Sandra Irish, Code 542 GSFC, Nov. 1987.
To illustrate the actual steps to do a computer analysis of an electric field, the modeling of the germanium detector
for the Gamma Ray Spectrometer (GRS) that flew on the Mars Observer is shown here. The ultimate objective of
this particular analysis was to find the optimum shape and “doping” of the germanium (Ge) detector to achieve a
uniform field within the germanium itself. The germanium is a solid cylinder about 56 mm long with a center hole
that extends about 7/8 down the length of the detector. The outer radius is 29 mm and the inner radius of the hole is
8 mm, as shown in Figure 6.1(b). It is first modeled as a 2-dimensional solid, (for practice; does not give the correct
answers), and then is modeled as a three-dimensional solid of revolution as required by the actual geometry. Then
the “doping” is added. This illustrates the progressive complexity of the modeling steps.
A. ANALYSIS
The analysis was divided into the following two parts: the two-dimensional (2-D) detector model and the three-di-
mensional (3-D) detector model. The 2-D model assumed the detector to be an infinitely long solid bar as shown in
Figure 6.1(a). The 3-D model, shown in Figure 6.1(b), represents a solid cylinder depicting the actual GRS detector.
The analysis was performed on the 2-D model first to obtain a better understanding of the analysis procedure required
prior to applying this procedure to the more complex 3-D detector model. Also, the cross-section of the infinite bar
model and the cross-section of the solid cylindrical model were the same, which simplified the development of the
3-D model.
Figure 6.2 shows a flow chart of the procedure used in the analysis to obtain a color graphics display of the voltage
and field strength distributions for the detector. Once the basic geometry of the detector was known, a model of the
detector was developed using PATRAN, a computer program which creates detailed finite element models efficiently.
Various modeling checks were performed on the elements to assure that accurate results would be obtained.
Chapter 6: Electrostatic Field Analysis by Computer
6-3
High Voltage Power Supply Design Guide for Space
Figure 6.1. 2-D and 3-D model representation.
Figure 6.2. Analysis procedure.
6-4
Chapter 6: Electrostatic Field Analysis by Computer
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
111 the 3-D model, a FORTRAN program was developed to reorder the grid points since the solid axisymmetric ele-
ments used in the model required a specific grid order format. Parameters, such as voltage boundary conditions and
charge density of the material, were then added to the finite element model to produce a model ready for submittal
to the NASTRAN Thermal Analyzer (NTA). The NTA was then used to perform a linear steady-state analysis with
HEAT SOL 24, NASTRAN HEAT = 1. This produced a voltage value for each grid point in the model and the voltage
gradients for each element in the model. A FORTRAN program was developed to extract the voltage gradient data
from the NTA output file and to produce the field strength distribution for the model. The field strength values were
calculated by obtaining the root sum of the squares of the voltage gradients.
The preferred method of reviewing the results from the analysis is by color graphics rather than scanning tables of
numbers. Therefore, the output from the FORTRAN program was organized into a format that was read by a MOVIE.
BYU, a computer program which displays results using color graphics. In addition, the Patran program can be used to
display the results. Color graphics plots of the voltage and field strength distributions were then created.
Two-Dimensional Detector Model
A MSC/NASTRAN model was developed for a cross-section of the infinitely long bar. Because of the symmetric
geometry of the cross-section and the symmetric loading that would be applied, only half of the section was modeled.
Figure 6.3 shows a plot of the 2-D NASTRAN model. The 2-D model contained 339 grid points and 296 CQUAD4
elements. A voltage boundary condition of 0.0 V at the inner edge and -4000.0 V on the outer edge was applied to the
model (refer to Fig. 6.4). The voltage distribution for the 2-D geometry was then calculated by the NTA. Figure 6.5
shows a color graphics plot of the equipotential lines, lines of constant voltage, calculated for the 2-D model. Figure
6.5 shows a fairly uniform distribution near the open end of the detector and shows that the lines become closer to-
gether and less uniform at the closed end. A field strength value for each element was then calculated from the voltage
gradients. Figure 6.6 shows the field lines, lines perpendicular to the equipotential lines, for the 2-D model. Also, field
strength values are labeled at various locations. The highest field strength of 688.7 V/mm was found at the lower inner
corner of the detector (where the equipotential lines are closest together), and the lowest field strength of 17.8 V/mm
was located at the outer bottom corner of the detector (where the equipotential lines are farthest apart).
Chapter 6: Electrostatic Field Analysis by Computer
6-5
High Voltage Power Supply Design Guide for Space
X-YVIEW ELEMENTS: ALL
ALPHA = 0.0 BETA = 0.0 GAMMA = 90.0 NU = 0.023
Figure 6.3. 2-D Model
6-6
Chapter 6: Electrostatic Field Analysis by Computer
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
OPEN END
CLOSED END
KEY:
ks^S^wafesI 0.0 volts
-4000.0 volts
Figure 6.4. GRS boundary conditions.
Chapter 6: Electrostatic Field Analysis by Computer
6-7
2-D Model
-2000.0 -4000.0
High Voltage Power Supply Design Guide for Space
o
d
Figure 6.5. Equipotential lines for the 2-D model.
6-8
Chapter 6: Electrostatic Field Analysis by Computer
-4000.0 volts
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
FIELD STRENGTH (VOLTS/MM)
190.7
190.5
190.3
190.9
190.5
190.0
192.5
190.8
188.5
199.8
191.8
181.8
239.2
154.9
Z X
Y
580.7
688
562.8
\ \\\<\\\
\ \ \\ \ wwr
\ \ \ \ \ \ \ \ \
■ V v \
76.9
450.4
184.4
17.8
X-YVIEW ELEMENTS: ALL
ALPHA = 0.0 BETA = 0.0 GAMMA = 90.0 NU = 0.030
Figure 6.6. Field lines of 2-dimensional model.
FIELD LINES
Chapter 6: Electrostatic Field Analysis by Computer
6-9
High Voltage Power Supply Design Guide for Space
Three-Dimensional Detector Model
The symetric geometry of the germanium detector made it possible to use MSC/NASTRAN’s axisymmetric elements
to create an appropriate 3-D finite element model. The CTRAPRG elements allows the user to model a trapezoidal
cross-section in the x-z plane. This cross-section is rotated about the z-axis to create a solid trapezoidal ring. In this
manner, the concept for converting the already existing 2-D GRS detector model to a 3-D model was relatively simple;
the CQUAD4 elements were converted to CTRAPRG elements.
However, there are a few restrictions on how the CTRAPRG can be specified. The grid points must be ordered counter-
clockwise, starting from the lower left. Also, all grid points must lie about the +x axis and to the right of the +z axis.
It was therefore necessary to pay close attention to the grid point connectivity, and use FORTRAN programs to ma-
nipulate both the connectivity and the grid point locations.
Preliminary results from the first 3-D model indicated that there were large electrical field strength gradients in the
x-direction. Because the field strength of two adjacent elements differed as much as 20%, a finer mesh was devel-
oped, increasing the model size from 296 to 1776 elements. This fine-mesh model is shown in Fig. 6.7. Several sets
of boundary conditions were applied to the model, with the final analysis configuration incorporating a voltage drop
across the detector of -4000 V (Fig. 6.4).
6-10
Chapter 6: Electrostatic Field Analysis by Computer
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
MARS OBSERVER, GRS
FINE MESH 3-D MODEL,
ELEMENT PLOT
Z X
Y
X-YVIEW ELEMENTS: ALL
ALPHA = 0.0 BETA = 0.0 GAMMA = 0.0 NU = 0.0
Figure 6.7. Fine mesh 3-D model, element plot.
Chapter 6: Electrostatic Field Analysis by Computer
6-11
High Voltage Power Supply Design Guide for Space
In order to determine whether the analysis was producing reasonable results, the computed field strength values were
compared to values calculated from the theoretical equation in polar coordinates, for the field strength in an infinitely
long cylinder (Fig. 6.8). A few specific locations near the open end of the detector were chosen for the NTA to theory
comparison because only the upper portion of the detector can be approximated by this theory. The comparison is
given in Table 6.2, which shows that the NTA results were within 1% of the theory. Also, a plot of the equipotential
lines is presented in Fig. 6.9. These lines were not uniformly distributed as the lines for the 2-D model (Fig. 6.5), but
instead were concentrated near the inner edge of the detector, as one would expect for cylindrical geometry, from
basic field theory, due to the Inverse Square Law.
Addition of Charge Density of the Material
Impurities in a germanium crystal lattice cause changes in the electric field, with the effective result being the addi-
tion of volumetric charge sources. This is often a desired effect which the detector designer can use to his advantage
through doping (controlled addition of impurities). The charge density distribution in the GRS detector varied linearly
along the z-axis, from 1.6 x 10 12 C/mm 3 at the open end to 2.4 x 10 12 C/mm 3 at the closed end. This was modeled in
the NTA by using the QVOL thermal load card, which added charge sources on every element.
The doped model was tested with the boundary conditions given in Fig. 6.4, and the electrical field strength results
agreed with the theory given in Fig. 6.8, with an error of less than 2%. Note that for this case, the doping was included
in the theoretical calculation and was a function of vertical (z) position. The results of the comparison between NAS-
TRAN and theory are summarized in Table 6.3, and the equipotential lines and field strength are displayed graphically
in Figs. 6.10 and 6.11, respectively. A maximum field strength of 935.3 V/mm occurred at the inside lower corner, and
a minimum of 21.4 V/mm occurred at the outside lower corner.
E(r)
pr
A V
1/4 | (r 2 2 - ^ 2 )
2z
l n ( r 2 / r<|) r
ln( r 2 / r 1 ) r
where:
p = charge source density (coulombs/mm 3 )
e = permitivity (farads/mm)
r = radial location
AV
r 2
r 1
= V 2 -V-|
= outer radius
= inner radius
Applies for r between ^ and r 2 .
Figure 6.8 Field strength in an infinitely long cylinder, from Ref. [11].
6-12
Chapter 6: Electrostatic Field Analysis by Computer
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Table 6.2. GRS field strength. Voltage drop = -4000 V. No doping.
Radius
mm
*E (NASTRAN)
V/mm
**E (Theory)
V/mm
%diff
10
310.1
310.6
-0.16
17
181.7
182.7
-0.55
29
107.0
107.1
-0.09
* Computed as an average of four elements near r, at a depth of 20 mm
** E(r)=\E(r)\
AV
In (r 2 / rjr
** J. Llacer, “Planar & Coaxial High Purity Germanium Radiation Detectors,” Nucl. Instr. Methods,
98(2), North Holland Publ., Jan. 1972.
Table 6.3. GRS field strength with doping. (Voltage drop of -4000.0 V, and doping was applied
with linear variation along axis of symmetry.
Radius
*E (NASTRAN)
**E (Theory)
%diff
mm
V/mm
V/mm
10
179.0
176.8
1.2
17
178.1
177.8
0.2
29
227.0
230.5
-1.5
*Computed as an average of four elements near r, at a depth of 20 mm
p r AV C-V)
2e ln(r 2 / r x )r e ln(r 2 / r x )r
where p = is the charge density at a depth of 20mm
e = permittivity (farads/mm)
r = radial location
AV = V 2 -V,
r 2 = outer radius
r = inner radius
** J. Llacer, “Planar & Coaxial High Purity Germanium Radiation Detectors,” Nucl. Instr. Methods,
98(2), Jan. 1972.
Chapter 6: Electrostatic Field Analysis by Computer
6-13
3-D Model
-2000.0 -4000.0
High Voltage Power Supply Design Guide for Space
o
d
Figure 6.9. Equipotential lines for 3-D model.
6-14
Chapter 6: Electrostatic Field Analysis by Computer
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
CD
T3
O
O
O
o
o
(M
O
d
o
o
<M
■
o
d
Figure 6.10. Equipotential lines with doping.
Chapter 6: Electrostatic Field Analysis by Computer
6-15
High Voltage Power Supply Design Guide for Space
CD
-o
o
O
CO
Figure 6.11. Field strength distribution with doping.
6-16
Chapter 6: Electrostatic Field Analysis by Computer
21.4 volts/mm
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Geometiy and Boundary Changes
Several refinements were made to the original detector model geometry to more closely match the actual design, and
to provide a feel for how geometry changes would affect the electric field strength and voltage distributions in the
detector. In detector design, sharp edges are sometimes rounded to reduce localized areas of high field strength and
remove areas of low strength. In the GRS detector, the two outside corners and the lower inside corner were rounded
slightly. This rounding was accomplished in the model by manually removing some elements on the outside corners,
adding a elements on the inside corner, and changing the shape of the elements that became the new edges. This
required the replacement of several CTRAPRG elements by CTRIARG elements. The CTRIARG is a solid axisym-
metric ring similar to the CTRAPRG element, but it has a triangular cross section instead of a trapezoidal one. The
restrictions discussed in an earlier section also apply to the CTRIARG specification. .The shape of the smoothed
corners was approximately circular, with radii of 3.5 mm on the lower outside corner, 4.0 mm on the inside corner,
and 4.1 mm on the upper outside corner. Figure 6.12 shows the rounded model geometry.
Two cases of the rounded model were analyzed. First, a model with rounding on only the inside and lower outside
corners was used with the original boundary conditions (Fig. 6.4). This resulted in only slight changes in the maxi-
mum and minimum field strength values compared to the fine-mesh 3-D model. Then, a model with all three corners
rounded was used with the outside boundary condition of -4000 V extended over the top of the model 9.8 mm from
the outside edge. The inside edge remained at 0 V. Doping, as discussed previously, was included in this analysis.
Figs. 6.13 and 6.14 show the resulting equipotential lines and field strength distributions in the detector.
One other variation of the original fine-mesh model was tested. The hole depth was changed from 49 mm to 31.65 mm
to determine how much this type of geometry change would lower the electrical field strength. The additional ele-
ments needed to close up the earlier deeper hole increased the model size to 1980 elements. This model is shown in
Fig. 6.15. The original boundary conditions (Fig. 6.4) and doping were used for this analysis. The maximum field
strength still occurred at the lower inside corner, but was reduced to 389.0 V/mm, or by about 58%. At the same
time, the field strength distribution was more uniform. A graphic representation of the field strength distribution is
given in Fig. 6.16.
III. Comments
Somewhat abbreviated versions of the computer program are available for desktop computers. An example is MAG-
GIE by McNeil-Schwendler Company. This can be run on IBM compatible desktop computers. Even so, the data-
inputting and manipulations are rather intricate and time-consuming. Also, most recently, programs called MAX-
WELL 2D and MAXWELL 3D, by ANSOFT Co. are available. Similar Electro -Magnetic Software can be obtained
from ANSYS or INFOLYDICA or MAGSOFT.
For many purposes, use of the Chapter 2, Table 2.2 formulae to obtain approximate maximum field strengths at the
surfaces of high voltage terminals and bus wires within a given power supply, combined with an estimate of average
field strengths as AV/d where d is the separation between a given terminal or wire and the nearest ground plane, is
all that is needed. The average field strength everywhere should be kept below 50 V/mil.
Chapter 6: Electrostatic Field Analysis by Computer
6-17
High Voltage Power Supply Design Guide for Space
MARS OBSERVER, GRS
ROUNDED MODEL,
ELEMENT PLOT
GRS
. \
A
/
. J
>
/
J"
>
X-YVIEW ELEMENTS: ALL
ALPHA = 0.0 BETA = 0.0 GAMMA = 0.0 NU = 0.0
Figure 6.12. Rounded model, element plot.
6-18
Chapter 6: Electrostatic Field Analysis by Computer
3-D Model
-2000.0 -4000.0
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 6.13. Equipotential lines for rounded model with doping.
Chapter 6: Electrostatic Field Analysis by Computer
6-19
High Voltage Power Supply Design Guide for Space
O
l
(/)
DC
CD
E
E
~o5
■4— 1
O
>
00
c\i
N
Figure 6.14. Field strength distribution for rounded model with doping.
6-20
Chapter 6: Electrostatic Field Analysis by Computer
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
MARS OBSERVER, GRS
31 .65 MM HOLE MODEL,
ELEMENT PLOT
Z X
X-YVIEW ELEMENTS: ALL
ALPHA = 0.0 BETA = 0.0 GAMMA = 0.0 NU = 0.0
Figure 6.15. The 31.65 mm hole model, element plot.
Chapter 6: Electrostatic Field Analysis by Computer
6-21
GRS - MO
389.0 volts/mm
High Voltage Power Supply Design Guide for Space
Figure 6.16. Field strength distribution for smaller hole and doping.
6-22
Chapter 6: Electrostatic Field Analysis by Computer
19.8 volts/mm
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
REFERENCES
1. Zienkiewicz, O.C., RL. Arlett, and A.K. Bahrani, “Solution of three-dimensional field problems by the Finite
Element Method,” The Engineer, 27 October 1967.
2. Silvester, P., and M.V.K. Chari, “Finite element solution of saturable magnetic field problems,” IEEE Trans., Vol.
PAS-89, pp. 1642-51, Oct. 1970.
3. Brauer, J.R., “Finite element analysis of electric fields using MSC/NASTRAN, Proc. Conf. Comp. Techniques
Electrostatic Fields, University of California at Santa Barbara, July 1978.
4. Kreith, F., Principles of Heat Transfer. 3rd ed., Intext Education Publ., New York, 1973, p. 93.
5. Brauer, J.R., “Electric Currents in Cathodic Protection Systems,” Proc. MS User’s Conf, Pasadena, California,
1979.
6. Brauer, J.R., “Saturated Magnetic Energy Functional for Finite Element Analysis of Electric Machines,”
C75-151-6, IEEE Winter Power Meeting, Jan. 1975.
7. Brauer, J.R., “Improvements in Finite Element Analysis of Magnetic Devices,” IEEE International Magnetics
Conf, Los Angeles, California, 1977.
8. Silvester, R, H. Cabayan, and B. Browne, “Efficient Techniques for Finite Element Analysis of Electric Machines,”
IEEE Trans., Vol. PAS-92, pp. 1274-81, July 1973.
9. Brauer, J.R., “Finite Element Analysis of Electromagnetic Induction in Transformers,” A77-122-5, IEEE Winter
Power Meeting, Feb. 1977.
10. Brauer, J.R., “Finite Element Analysis of Electric and Magnetic Fields,” High Voltage Workshop, Clarksburg,
Maryland, October 1979.
11. Llacer, J., “Planar and Coaxial High Purity Germanium Radiation Detectors,” Nucl. Instr. Methods, 98(2),
Jan. 1972, North Holland Publ.
Chapter 6: Electrostatic Field Analysis by Computer
6-23
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 7. Design Examples of Successful High Voltage Power Supplies
Introduction
The creation of high voltage power supplies, or any other high voltage assembly or device falls, broadly speaking,
into two stages:
(1) the electrical/electronic design, and
(2) lay-out/packaging/insulation build-up.
The two stages are, of course, intertwined, and both have to be kept in mind throughout the theoretical planning and
the practical hardware fabrication. The emphasis of this book is mostly on stage (2), therefore, photographs and some
comments on the manufacturing of several high voltage power supplies are presented. It is also deemed worthwhile
to present circuit schematics and summaries of circuit analyses of one or two supplies. The latter is a concession to
stage (1) mentioned above; however, a step-by-step development of electronic design is not intended to be done in this
book.
Operational guidance for test and flight of each high voltage power supply is also very important. Therefore, a single
example of one such set of instructions is included as an appendix to Chapter 7, for the ACD/GLAST High Voltage
Bias Supplies (HVBS).
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-1
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE A
A -25 kV supply for the High Resolution Spectrograph (HRS) and Faint Object Spectrograph (FOS) of the Hubble
Space Telescope Project by Joseph A. Gillis, GSFC. (Later adapted by Martin Marietta Co., of Denver, Colorado,
and by Ball Brothers of Boulder, Colorado.)
Specification for a Digitally Controlled HVPS for the HRS Digicon
1. Method of Control: A ten bit digital word will control the voltage output of the device from 0 to -24 kV. This word
may be presented to the supply either serially or in parallel, and in accordance with the interface provisions of the
HST (Hubble Space Telescope) “Science Instruments to Control and Data Handling Subsystem Interface Require-
ments Document” (S-725-4). The eight least significant bits will control the supply’s output from -18 kV to -24
kV to the specifications listed below, while the two most significant bits will permit a stepped turn-on from 0 to
-18 kV with performance to specifications not being mandatory. Accuracy and linearity of the control word to
output voltage function shall be ±1 LSB over the -18 to -24 kV range.
2. Voltage Output: 0 to -24 kV (Turn-on Procedure), -18 to -24 kV (Normal Operation)
3. Voltage Regulation: ±0.05% over line, constant load; that is ±9 V to ±12 V.
4. Voltage Ripple: ±1 V p-p
5 Output Current: 60 pA at 24 kV; this includes 40 pA load and 20 pA feedback.
6 Input Voltage: 19 V ±2%
7. Input Power: Less than 3 W
8. Temperature Range: 0°C to 40°C Operating; -20°C to 60°C storage
9. On/Off Command: Relay controlled switching in accordance with HST interface specifications; that is about
1.25 W.
10. Monitor/TM Functions
Analog: (a) 0 to 5.10 V tracking; 0 to 24 kV output
(b) 0 to 5.10 V tracking; -18 to -24 kV output (suppressed zero)
(c) 0 to 5.10 tracking, full range input current
Digital: (a) 1-bit On/Off monitor, from power switching relay
(b) 10-bit status of D/A holding registers
11. Other Features: “Soft” turn-on; output voltage will require at least 2 s to reach 1% of final value.
7-2
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.A.I. Circuit for FIRS power supply, developmental model.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-3
5. Input Connector: Cannon 50 pin female
6. Relay: Potter Brumfield TL17DA12
High Voltage Power Supply Design Guide for Space
Operation of -25 kV, FOS Supply
FOS and HRS High Voltage Power Supply Circuit Operation, -25 kV
The power supply operates from a +19 V source. Ql, Q2, and associated circuitry form a current-fed push-pull sine
wave oscillator whose frequency of operation is determined by the inductance of Tl, capacitors C2 and C3, and to a
lesser extent, the effects of the following circuitry. The peak voltage output from Tl is approximately 160 V.
T2 is used to multiply the output of Tl and applies this to the input of the multiplier stack. The input voltage level to
T2 will be the output of Tl minus the drop across Q3, which is the regulating transistor. The maximum output of T2
is approximately 1400 V peak.
The multiplier stack is a series-parallel connected 10 stage device, which boosts the T2 output to a maximum of about
-25,000 V. The output of the multiplier stack is RC filtered and brought out of the unit by high voltage shielded cable
rated at 30,000 V. The shield is kept at chassis ground. A load return line is also brought out at this point to ensure
that the HV load current returns directly to the multiplier stack output.
The output voltage is sensed by voltage divider R6-R14 and Ul. U1 nulls at zero and produces an output that is of op-
posite polarity and about l/5000th of the HV output. The output of Ul is fed to U2, which nulls with the commanded
reference voltage from U3. The output from U2 drives regulating transistor Q3.
Three monitor voltages are provided to check on the state of operation of the unit. U4 provides a voltage that follows the
output of Ul. U5 produces a voltage that goes from approximately 0 V to +5.1 V as the HV output goes from -18,000
to -24,000 V. U5 produces a voltage that follows the output voltage of T3. T3 senses the emitter currents of Ql and
Q2 and converts this to a voltage across R49. The state of the HV on-off relay can also be monitored. The resistance
to ground of pin 44 will be 0 Q when K1 is closed and infinite ohms when K1 is open.
R34, which makes up part of the voltage divider from U3, is a Texas Instrument sensistor (570 Q) used to keep the
HV output within specification over the temperature range of 0°C to +40°C.
All the high voltage areas (within the dashed line on the schematic) are encapsulated in CONAP EN-11.
R15 (100 kQ) is placed across Q3 to ensure that the collector of Q3 stays below approximately 200 V. It was found
that the collector voltage of Q3 tends to go to the peak-to-peak output of Tl when the unit is commanded to low levels
(<700 V). Q3 is not within the encapsulation and the unit will be operated in vacuum. As a result of this, the minimum
output voltage of the unit is approximately -500 V at room temperature. Other than for this, the unit is commandable
up to -24,000 V in 1024 steps.
Circuit operating frequency is at 35 kHz cycles. Input current is 125 mA maximum.
The circuit diagram in Fig. 7. AT is essentially this:
Sine wave Oscillator -^Transformer-* Bridge Regulator -* High Voltage Transformer -*
High Voltage Multiplier
The Bridge Regulator in the middle achieves isolation of the high voltage ground from power ground.
7-4
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
V s = 1 50V, 0 - pK v s - V R = 1 50V, 0 - pK
3000 Vqut
Figure 7.A.2. Bridge regulator.
This Bridge Regulator has the advantage that V R on the transistor is relatively low, even for output voltage greater
than 10 kV, and thus, it was commonly used for ground isolation up to about 1983. However, the input current can
increase dramatically if the operating frequency is different from the resonant frequency of the 1:30 turn transformer.
This difference can be caused inadvertently by stray capacitances of coaxial cables used or by changes in temperature
cycling. So this scheme is not used much any more.
Also, in the High Voltage Multiplier:
V
Actual
V
r Ideal
where N= the number of stages in the Multiplier. One can put the stages in a single multiplier in groups of five in
“parallel” to reduce losses and ripple. Then, however, the input capacitor to the multiplier stacks will need a rating
commensurate with the output voltage. The transformer itself should be able to withstand full output voltage in the
event of a shorted output.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-5
High Voltage Power Supply Design Guide for Space
Figure 7.A.3. Photograph of the HV portion of the HRS power supply before potting.
7-6
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Discussion of Fig. 7. A. 3:
Several things are noteworthy in the photograph showing the construction of the high voltage portion of the power
supply before potting:
1. This is an example of a high voltage portion that was later solid potted with Conathane EN-11, including
the high voltage transformer, T2. The potting extended up to the top of the shielding box. (This is actually
undesirable. The top of potting should be a mechanically unconstrained surface.)
2. The careful construction of the feedthru(s) through the shielding walls should be noted.
3. Hardwiring is done with bare buswire; the later EN-11 potting serving as insulation.
4. Stress relief bends and loops are on each part.
5. Use is made of three Custom Electronics Co. impregnated micapaper capacitors, 1000 pF, 37.5 kV rated.
6. Caddock type MG 750 resistors are used.
7. There is a problem site : Note where the high voltage cable exists from inside the box. Just before the exit
there is an anti-corona donut attached to the cable where the shielding is peeled back. There was prolonged
(about 6 years) storage after the Build and the Thermal Vacuum test before Flight. Failure occurred during
the Thermal Vacuum test at the braid peel-back spot and at the anti-corona donut:
(a) The EN-11 had cracked at the donut, probably induced by the differential expansion/contraction be-
tween the large metallic donut and the EN-11, probably further exacerbated by the lid pressing on the
top surface of the EN-11 potting.
(b) Black residue was seen upon cutting into the potting, where the shielding braid was bent back over the neck
of the anti-corona donut.
It is believed that the problem was solved by deleting the too large anti-corona donut and using a small anti-corona
ball at the braid peel-back area. This is a good example where an optimum electric field design in theory clashes
with differential thermal expansion properties of the different materials involved in the actual construction, and an
engineering compromise has to be reached. The theoretical design was published before the aforementioned failure
was discovered*
*High Voltage Power Electronics Packaging on NASA’s Space Telescope, by, S.R. Yadavalli, J.L. Westrom, and J.W. Williams; 17th
IECEC, 1982, p. 211.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-7
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE B
A -15 kV Plasma Experiment Power Supply forIMP-G, by, John L. Westrom, Goddard Space Flight Center.
This -15 kV Plasma Experiment power supply for IMP-G is an earlier supply, already described in Sutton and Stern’s
NASA-TN-D- 7948 Spacecraft Power Supply Construction book, and also in GSFC document X-7 16-68-252. It is briefly
taken up here again because it is one of very few examples of solid potting with RTV-615 silicone rubber. Although
in the last 20 years, mostly polyurethanes have been used for solid potting, nevertheless there recently (year 2004)
were requests to GSFC for advice on how to make solid potting with silicone rubber a success.
The -15 kV supply was “located immediately adjacent to the instrument to which it gave the high voltage bias, thus
eliminating the need for HV cables and connectors exiting from the RTV-615.” For the flight units, the HV output
lead in the center of the top of the circuit board was not the insulated wire shown in the engineering model in Fig.
7.B.I. The flight model had a bare wire helix through the encapsulant through a Kel-F insulated circular cover on
top. This cover was directly above (about 1 in) the top side of the assembled circuit board and held a smooth corona
ball outside and above the Kel-F barrier (Fig. 7.B.2). This barrier extended only a fraction of an inch down along
the cylindrical side of the potted supply by means of a circular metal ring. The potting on the circular bottom of the
assembly was only about 0.25-in thick because it had no high voltage points on it. The diameter of the power supply
was approximately 9.5 cm.
All of the HV components were mounted on etched Teflon stand-offs on the top side, and the Kel-F circular disc was
especially etched also. The solid disc ceramic capacitors were not wax-impregnated, but especially cleaned and then
primed with a 50-50 mix of Epon 828/Versamid 140 and overcoated with silicone rubber DC 93-500. Probably the
same treatment was done to the other HV circuit parts; this adhered well to the RTV-615 encapsulant.
The vacuum encapsulation was carried out for the entire assembly with RTV-615 in a removable mold of slightly smaller
internal dimensions than the metal shielding chassis. The cured, potted power supply was then removed from the mold.
It was then coated with a thin layer of a mix of RTV-615 and powdered carbon black, on the cylindrical and bottom
surfaces, so that bleeding off of static charge to the inside of the grounded chassis box could occur. The tolerances of
dimensions between the outside of the coated power supply and the inside of the grounded metal chassis box are not
available to the writer any more. But it has to be such that during thermal cycling the RTV-615 silicone rubber can
expand and contract with respect to the metal shielding chassis without being constrained. No adhesion is necessary
of the potting to the inside of the grounded chassis box, if bleed-off of charge is assured by the carbon-containing
coating. Thus, if constraining of the silicone rubber is avoided during its large thermal expansion and contraction
upon thermal cycling, then the silicone rubber will not crack. The cracking and tearing loose from circuit components
of silicone rubber during thermal cycling must be avoided by taking its large thermal expansion and contraction into
consideration in the original design.
7-8
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.B.I. High voltage supply component layout, top.
Figure 7.B.2. A 15 kV supply, top.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-9
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE C
A Quick Response, 4 Decade Logarithmic High Voltage Stepping Supply, to ±2500 V (also called HAPI LAP!), by
Henry Doong, GSFC.
Packaging
The High Voltage portion of this circuit, namely the HV shunt regulators and the HV generator/modulator, were only
conformally coated with EN-11 polyurethane, not solid potted. Because the maximum voltages were only ±2500 V,
this packaging was successful. The coating was applied by spraying, not brushing. The parts were spot-staked. For the
techniques of coating for high voltage, see Materials Processing Document S-313-019, NASA/GSFC: “Conformal Coat-
ing of HV Printed Wiring Boards with a Room-Temperature Curing Two-Part Urethane Resin,” September 1992.
Circuit Design
The circuit discussion of this power supply is best accomplished with a document by the design engineer himself,
which is herewith included. The Figure numbers of this document are left as in the original document
INTRODUCTION
The design of low power, high voltage supplies usually consists of an amplitude modulated high frequency tuned
oscillator in conjunction with a voltage multiplier and filter to raise the available voltage to the desired output level.
This type of circuit arrangement does not meet the design goals when a fast response time is required as the output
voltage response time is prolonged by the time constant associated with the voltage multiplier-filter capacitor and
load. To compensate for the fast response time, either or both the capacitor value and the load resistance must be
reduced, but a reduction in resistance leads to a large increase in input power while a smaller capacitor value results
in a large ripple voltage; or if operated at higher frequencies, the ripple voltage improves but the efficiency is reduced.
To minimize the response time without increasing the input power or output ripple voltage, the high-voltage stepping
supply uses an active high-voltage shunt regulator at the high-voltage output.
To meet the design goals of the four decade dynamic range, two innovative circuits are used. One circuit uses the 6 bit
to 32 step decoder, which requires only 12 scaling resistors rather than the 32 in a conventional design. The other circuit
has two complementary amplifier channels with each channel producing one-half of the 64 step staircase waveform;
each channel is “gated” on at the appropriate time to produce a complete staircase voltage at the output.
The high-voltage stepping supply is equipped with a self-contained low voltage source for the internal circuits, an
overload and over voltage protection circuit, and two output voltage monitor circuits. The supply, including the supply
housing, weighs 1-1/2 pounds and occupies a volume of approximately 63-cubic inches.
SYSTEM DESCRIPTION
The stepping supply functional block diagram, Fig. 1 [7.C.1], consists of the following units.
• 6 bit to 32 step logarithmic staircase generator (two consecutive 32 steps per cycle)
• Reference voltage source
• Signal processor
• Oscillator, modulator, and parallel charge multipliers
• High voltage DC shunt regulators
7-10
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
OPERATION
Six Bit to 32 Step Logarithmic Staircase Generator
The 6 bit to 32-step logarithmic staircase generator, Fig. 2 [7.C.2], accepts the 6 bit binary input command at the ba-
sic clock rate of 64 steps per second (15.6 ms step dwell time). Each bit is passed through a buffer stage (Ql through
Q 6 ) and an inverter logic IC-1 (CD4049A). The three most significant bits (MSB) interface with the multiplexer IC-2
(CD4051A). The other three least significant bits (LSB) are level shifted by Q 7 , Q 8 and Q 9 . Thus for the LSB, a logic “1”
is zero volt and a logic “0” is -10 volts. The level shifters are necessary to interface with multiplexer IC-3 (CD4051 A)
as the switches are connected to the negative potential.
The two multiplexers, IC-2 and IC-3, combined with amplifier A1 and the 12 resistor matrix, constitute a multigain
summing amplifier, with resistors R f as the feedback resistor and R in as the input resistor. Each switch in the multi-
plexers is selectively turned on by the 6 bit input code. Thus for any step, the amplifier A1 output voltage El can be
written as:
K
(i)
Where:
E,= the output voltage of A t
Rf= feedback resistance selected by LSB code
R in = input resistance selected by MSB code
V r = reference voltage, 7.50 volts
The circuitry provides 32 logarithmically scaled voltage levels at the output E[. Each level is determined by two dis-
crete switch closures. The MSB code closes a discrete switch in IC-2 and the LSB code closes another discrete switch
in IC-3. Accordingly, one scaling resistor of four is chosen for R in and one scaling resistor of eight is chosen for R f .
The total number of combinations available is 32. Since the basic clock rate is 64 steps per second, the logarithmic
staircase will be generated twice per second. The resistive scale factors are chosen to cover two decades of voltage
over the 32 levels. Table 1 [7.C.1] shows amplifier A, output E! vs. input command code and the corresponding scal-
ing resistors.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-11
High Voltage Power Supply Design Guide for Space
Table 1. [7.C.1]. Six bit to 32-step voltage levels and scaling resistors versus command code.
No.
Code* Ei
Rin
Rf
No.
Code*
El
Rin
Rf
1
X00000 -7.500
100K
100K
17
X10000
-0.7500
1M
100K
2
X00001 -6.495
100K
86.6K
18
X10001
-0.6495
1M
86.6K
3
X00010 -5.625
100K
75. OK
19
X10010
-0.5625
1M
75. OK
4
X00011 -4.872
100K
64.9K
20
X10011
-0.4872
1M
64.9K
5
X00100 -4.218
100K
56.2K
21
X10100
-0.4218
1M
56.2K
6
X00101 -3.654
100K
48. 7K
22
X10101
-0.3654
1M
48.7K
7
X00110 -3.165
100K
42.2K
23
X10110
-0.3165
1M
42.2K
8
X00111 -2.739
100K
36.5K
24
X10111
-0.2739
1M
36.5K
9
X01000 -2.374
316K
100K
25
X11000
-0.2374
3.16M
100K
10
X01001 -2.056
316K
86.6K
26
X11001
-0.2056
3.16M
86.6K
11
X01010 -1.781
316K
75. OK
27
X11010
-0.1781
3.16M
75. OK
12
X01011 -1.542
316K
64.9K
28
X11011
-0.1542
3.16M
64.9K
13
X01100 -1.336
316K
56.2K
29
X11100
-0.1336
3.16M
56.2K
14
X01101 -1.157
316K
48. 7K
30
X11101
-0.1157
3.16M
48.7K
15
X01110 -1.002
316K
42.2K
31
X11110
-0.1002
3.16M
42.2K
16
X0 1111 -0.8679
316K
36.5K
32
Xlllll
-0.0868
3.16M
36.5K
* “X” denotes a “don’t care bit.”
REFERENCE VOLTAGE SOURCE
The reference voltage V„ Figure 2 [7.C.2], for the staircase generator is obtained from two stable 6.8 volts, temperature-
compensated zener diodes, D, and D 2 . The field effect transistor FET (Q 10 ) is employed to furnish a constant current
source to the two diodes. Resistors R, and R 2 scale the zener voltage to exactly 7.50 volts. Amplifier A 2 serves as the
buffer for the low source impedance to the load.
SIGNAL PROCESSOR
The signal processor, Figure 3 [7.C.3], has four basic functions:
1. to condition staircase waveform Ej so that the proper polarity and format are provided at V p and V n
2. to amplify and deliver a pair of mirror image staircase voltages at +LV and -LV
3. to compensate for the temperature and tracking error of each mirror image voltage
4. to alternately select each of the two amplifier channel voltages which produce +V 0 and -V 0 outputs.
As shown in Figures 2 [7.C.2] and 3 [7.C.3], the 32 level staircase E, waveform is applied to each of the two sets of
amplifiers, One set, consists of two inverting unity gain amplifiers A, and A 2 . When the 2 bit logic is “Zero,” ampli-
fiers A! and A 2 produce a pair of mirror image staircase voltages V p and V n , refer to Figure 4 [7.C.4], to set the highest
two decade voltage outputs in the first half cycle of the total staircase waveform. However, when the 2 5 bit logic is
“One” during the second half cycle of the total staircase waveform, the voltage E! is blocked from going through the
amplifiers. The voltage at V p becomes slightly positive while the voltage at V„ becomes slightly negative.
V p is used to dynamically control the positive shunt regulator to provide the regulation output +V 0 during this half cycle
(steps 1-32). V p is also compared with the feedback voltage from the multiplier stack output +HV. The difference is
the error voltage and is amplified to control the +HV level via the modulator.
7-12
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
V n , the mirror image of V p , controls the negative shunt regulator output -V 0 during the same first half cycle to provide
the regulated output -V 0 .
The second amplifier set is used to provide the logarithmically scaled voltage levels during the second half cycle
(steps 33-64). It consists of amplifier A 3 (gain = -3.33) and amplifier A 4 (gain = -1) and a temperature compensation
network R, - D b refer to Figure 3 [7.C.3].
The staircase outputs, +LV from amplifier A 3 and the mirror image -LV from amplifier A 4 , refer to Figure 4 [7.C.4],
are used directly as outputs +V 0 and -V 0 respectively during the second half cycle, (steps 33-64). Actually these out-
puts are connected to +V 0 and -V 0 by means of the high voltage steering diodes, D 10 and D 20 , refer to Figure 6 [7.C.6],
During the first half cycle (steps 1-32), when the highest two decades of output levels are regulated by the high voltage
shunt regulators, both of these diodes (D l0 and D 20 ) are reverse biased since ±LV outputs never exceed ±V 0 .
During the second half cycle, the slight positive voltage on V p is sufficient to cut off the parallel charge multipliers;
therefore, no voltage is generated at either +FIV or -HV output terminals. This allows the two steering diodes D 10 and
D 20 in Figure 6 [7.C.6] to conduct, providing the lower two decade outputs to +V 0 and -V 0 . The voltage drop across
each diode is compensated by an equal offset diode (D,) voltage (V d ) to the amplifier A 3 input in Figure 3 [7.C.3],
Diode D is similar to the steering diodes, and resistor R, provides the necessary compensating current.
The equations for defining the outputs: V p , V n , +LV, and -LV in Figure 3 [7.C.3] are:
First half cycle ( steps 1-32)
V P = (-!)(-!) E,=E,
V n = (-1) E 1
+LV= (-3.33) E, + (-1) (V d )
-LV=(-1)(+LV)
Second half cycle (steps 33-64)
V p = (— 1)(— 1)(“1”)= positive voltage greater than 1 volt
V n = (-1)(“1”)= negative voltage greater than 1 volt
+LV=(-3.33)E 1 + (-l)(V d )
-LV=(-1)(+LV)
( 2 )
(3)
(4)
(5)
( 6 )
(7)
( 8 )
(9)
Notes:
E! and V d are always negative
(“1”) denotes 2 5 bit logic “One”. It is always greater than E) in magnitude and positive.
Using -7.5 volts as the maximum voltage for E[ (steps 1 or 33), the following voltages yield:
First half-cycle, step 1
(From eq. 2) V p =-7.5 volts
(From eq. 3) V p =+7.5 volts
(From eq. 4) +LV=25 + 3 = 28 volts (assume V d = -3 volts)
(From eq. 5) -LV= -28 volts
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-13
High Voltage Power Supply Design Guide for Space
Second half-cycle , step 33
(From eq. 6) V p = positive voltage greater than 1 volt
(From eq. 7) V n = negative voltage greater than 1 volt
(From eq. 8) +LV = 25 + 3= 28 volts (assume V d = -3 volts)
(From eq. 9) -LV = -28 volts
OSCILLATOR
The 60 kHz sine wave oscillator, Figure 5 [7.C.5], consists of transistors Qi and Q 2 , inductor L and transformer T,,
refer to Reference 2. The operating frequency is determined by the secondary inductance and shunt capacitance C,,
Inductor L keeps the current through it fairly constant at all times, reducing sharp current spikes when the transistor
is turned on and minimizing conducted radiated EMI to nearby circuits.
The peak primary voltage (jt/ 2 times V cc ) is 57 percent higher than the standard Hartley or Colpitts oscillators. When V cc
equals 24 volts and the transformer T, step-up ratio is 3, the secondary peak output voltage of 226 volts is realized.
MODULATOR
The diode modulator is a full wave diode bridge (D, through D 4 ), Figure 5 [7.C.5]. The bridge circuit is connected
in series with transformer T 2 primary, with transistors Q 3 and Q 4 acting as the variable load across the bridge arms.
Transistors Q 3 and Q 4 are in parallel to share the worst case load dissipation condition.
The oscillator output from the step up Ti secondary winding is divided between the load and transformer T 2 . The
voltage across the secondary winding is again stepped up to produce about 1 kV peak-to-peak. Transformer T 2 is
tuned to the same frequency as the oscillator frequency with capacitor C 2 so the minimum reactive load is reflected
to the oscillator and modulator circuitry.
PARALLEL CHARGE MULTIPLIERS
The relatively high AC voltage (approx. 1000 Vpp) from transformer T 2 secondary is fed into two sets of multiplier
stacks, Figure 5 [7.C.5]. One set generates a positive high-voltage staircase at +HV and the other set produces a negative
high-voltage staircase at -HV. Each stack, composed of 6 diodes and 6 capacitors, is connected in a parallel-charge
configuration rather than in a series-charge configuration as in the Cockcroft-Walton multiplier. The parallel-charge
configuration is more efficient when several voltage multiplier stages are used. The trade-off is the AC capacitor peak
inverse voltage must be N-times larger in the N-th stage of the multiplier.
The positive high-voltage +HV and negative high-voltage -HV are the result of the rectified multiplier stack voltages.
Their magnitude is dictated by the inputs to the summing amplifier A b Fig. 5 [7.C.5], namely the staircase waveform
V p and the fixed offset voltage. As each step advances, the staircase waveform at V p immediately introduces an error
signal to the amplifier A! input. The amplified error signal drives the base terminal of the two parallel transistors, Q 3
and Q 4 , connected across the diode bridge. This drive signal determines the available collector to emitter resistance
across the diode bridge. Since the oscillator signal is shared between the diode bridge and the primary winding of
transformer T 2 only that portion across T 2 is stepped up and multiplied. The multiplier output voltage +HV adjusts
accordingly until the feedback current through resistor R f cancels the error signal that has been generated by the V p
staircase waveform.
Thus summing amplifier A b combined with the diode modulator and the parallel charge multiplier, Fig. 5 [7.C.5],
may be considered as an operational amplifier with a gain equal to:
+HV=-V P R f /R in + (V s ) R f /R b
( 10 )
7-14
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
where
R f = 110 MQ,
R in = 301 kQ,
R b = 3.3 MQ (offset input resistance), and
V s = 3 V (offset voltage).
Therefore, +HV = -365.4 V p + 100.
As mentioned earlier V p in step 1 is -7.50 volts, substituting this voltage in equation 10 yields +HV equal to 2840 volts.
This is the maximum voltage developed at +HV, with the same negative voltage occurring simultaneously at the -HV
terminal.
Normally, the +HV and -HV outputs can be used as the final output voltage for the external load if the step time is
at least several times larger than the transient time. The transient time is the rise or fall time of the RC time constant
of the multiplier stack used and is typically about 20 milliseconds. When the step time is small, as in this application
(15.6 ms/step), several alternatives can be used to meet the requirement. Some examples follow below.
Reduce the Load Resistance
Assuming that the equivalent capacitance of the multiplier stack is 1000 pF, to obtain a time constant of 1 millisecond
a 10 6 ohm resistor must be used from the +HV output to ground. The 10 6 ohm resistor will dissipate 6.25 watts of
power at +HV when the +HV equals 2500 volts. A similar resistor must be used between the -HV output and ground
which dissipates an additional 6.25 watts.
Decrease the Multiplier Capacitance
Assuming the load resistance is 10 8 ohms, to keep the power dissipation to a reasonable level for the same time constant
of 1 millisecond it would require the equivalent capacitance on the multiplier to be no larger than 10 pF. A multiplier
stack with such small capacitors in the chain would be inefficient with a large ripple voltage that may exceed the
magnitude of the step voltage itself. Although the ripple voltage may be reduced if the oscillator frequency is near
one MHz, but operating at this frequency would create many new problems such as: diode leakage in the multiplier,
RF shielding, and excessive core loss in the transformer, etc.
Add an Active Circuit to the Load
By adding a dynamic dc shunt regulator to the load, the low resistance shunt would discharge the excessive charge
during transition times but then would act as a very high resistance during the steady state condition. This shunt
regulator is described in the following section.
HIGH-VOLTAGE DC SHUNT REGULATOR
The use of two shunt regulators, one to regulate the positive voltage output from +HV and the other to regulate the
negative voltage output from -HV, are coupled through isolation resistors R d as shown in Figure 6 [7.C.6]. The resistors
allow the regulated outputs +V 0 and -V 0 to track the input waveform V p faithfully with a fixed gain; Figure 7 [7.C.7]
shows the improvements to these waveforms. The transition time between steps of the +HV waveform is measured to
be about 20 milliseconds. For high-level steps, the shunt regulator output +V 0 is less than 1 millisecond. For low-level
steps, the transition time for both +LV and +V 0 are approximately 100 microseconds.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-15
High Voltage Power Supply Design Guide for Space
Operation of the DC Shunt Regulator
Amplifier A l5 Figure 6 [7.C.6], receives waveform V p via input R connected between V p and the virtual ground input.
It also receives the stepping supply output +V 0 as a feedback signal via resistor R 2 . The shunt regulator and amplifier
Ai, when combined, have a negative overall gain determined by the ratio of R 2 to R,. The voltage +V 0 is then
+V n = ^V =-333 V ■
K ' (ID
Compare equation 11 with equation 10; (+HV) - (+V 0 )= 0.1 V 0 + 100 or the difference of 100 volts plus 10 percent
of the output voltage. The difference is the drop across the resistor R d which provides a margin of regulation for the
shunt regulator.
Because of the relatively high maximum voltage (2500 volts) appearing at +V 0 and the limited breakdown voltage of
available solid-state devices, the shunt regulator, Reference 3, is composed of 9 stages connected in a totem pole con-
figuration. The NPN transistors (Qi to Q 9 ) are rated at 400 volts, and the zener diodes (D to D 9 ) are rated at 300 volts.
This arrangement permits the voltage +V 0 to be divided across the 9 transistors.
Each shunting zener diode limits the voltage across its companion collector to emitter junction to about 300 volts.
The emitter and base of each transistor are connected by a separate resistor for shunting transistor leakage current
in a conventional manner. All bases are coupled to amplifier A! output via separate high-voltage diodes (D u to D 19 )
in series with current limiting resistors (R 5 to FT , ,). These diodes are reverse biased whenever their connecting base-
to-emitter junction resides at 300 volts or greater. Amplifier A, output can drive each base (Q to Q 9 ), one at time,
into active linear conduction. If the base is overdriven, the transistor saturates and pulls the next series transistor into
conduction.
The composite shunt regulator is variable driven into conduction by the output of amplifier A> to a voltage at terminal
+V 0 , refer to equation 11. Current conducts from +HV, through dropping resistor R d , into the shunt regulator. This
current at +V 0 passes through all zener diodes whose companion transistors are biased off. It then passes through the
transistors to the -10 volt return. For discussion purposes, it is assumed that that the zener diodes have a breakdown
voltage of 300 volts and that the instantaneous voltage of +V 0 is 1400 volts. Starting from terminal +V 0 , the first four
zener diodes (D, to D 4 ) are in conduction and provide a voltage across diodes D, to D 4 of 1200 volts. The remaining
diodes (D 5 to D 9 ) and the first four transistors (Q, to Q 4 ) are at cutoff. The conduction path then passes through the
fifth until the ninth transistor (Q 5 to Q 9 ). Q 5 is dynamically controlled by the output of amplifier A! so that it has a
collector-emitter voltage drop of approximately 200 volts. Transistors Q 6 to Q 9 are saturated and have a voltage drop
approaching zero volts producing the 1400 volts from terminal +V 0 to ground. The first four transistors (Q, to Q 4 ) are
maintained at cutoff because the emitter voltage established by the first four zener diodes are higher than the small
base voltage established by amplifier A,.
Thus for any instantaneous voltage +V 0 , not more than one of the transistors is dynamically controlled by amplifier
A[. All transistors above the controlled transistor are at cutoff and the transistors below the controlled transistor are
saturated. Each transistor assumes control in different adjoining 300-volt ranges, whereby transistor Q dynamically
controls +V 0 between zero and 300 volts, transistor Q 2 dynamically controls +V 0 between 300 volts and 600 volts,
and so forth. The shunt regulator is returned to a -10 volt source rather than to ground in order to enable the low
gain amplifier output +LV to pass through the steering diode D 10 to the output +V 0 , when +V 0 is less than 25 volts.
As mentioned before, when the output +V 0 is between the range 2500 volts and 25 volts, it obtains its voltage from
the parallel charge multiplier output +HV through the dropping resistor R d . When it is in the 25 volt or less range, it
obtains its voltage from the low gain amplifier output +LV through the steering diode D ]0 . The control for this process
is a logic “OR” function. The inputs to the “OR” are +HV and +LV and the output is the +V 0 voltage. For the MSB
range, a staircase voltage at E! (Figure 1 [7.C.1]) is amplified by a gain of -333 to become +V 0 by controlling the paral-
7-16
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
lei charge multiplier voltage +HV and the dc shunt regulator gain. For the LSB range this amplification gain is only
-3.33 by controlling the low-gain amplifier. Therefore, E[ = -7.5 volts, +V 0 would be 2500 volts via the high channel
and +25 volts via the low gain channel. The LSB range covers the second half cycle (steps 33 to 64) of the 64 step
cycle. During this period, the +V 0 output is obtained via the low gain channel and is equal to:
+V 0 =+LV-steering diode drop (12)
=28 - 3 = 25 volts.
Also during this period all the transistors in the shunt regulator are saturated on and the diode modulator is turned
off, effectively blocking any voltage build up at +HV. This condition is created by the 2 5 bit that causes a slight posi-
tive voltage at V p . The shunt regulator is dormant during this period with a passive load from +V 0 to -10 volts of
100 kilohms.
The negative voltage shunt regulator operates similar to the positive voltage shunt regulator. There two additional
circuits used in the negative voltage regulator to bring its voltage -V 0 exactly equal to, but opposite in polarity from the
+V 0 voltage. One is the “BAL. ADJ.” trimming resistor coupled to the input of the amplifier A 2 in Figure 6 [7.C.6]; this
adjustment is for the high gain channel. The other is the low gain adjust (LGA) resistor network coupled to the input
of amplifier A 4 , Figure 3 [7.C.3], to adjust for the low gain channel output. This slight difference in voltage between
the two outputs is due to component tolerance error in each channel. A to be determined (TBD) capacitor across R 2
in the feedback loop of each shunt regulator is used to adjust for critical damping in the operational amplifier loop.
CONCLUSION
Although the high-voltage stepping supply described here pertains to step-down staircase voltages, a step-up staircase
voltage may be constructed by using the same design principles.
Chapter 7: Design Examples of Some FIigh Voltage Power Supplies
7-17
High Voltage Power Supply Design Guide for Space
o
03
CD
03
CD
03
O)
O
CD
CM
CO
CD
CO
CM CM CM CM CM CM
Figure 1 [7.C.1]. Stepping supply functional diagram.
7-18
Chapter 7: Design Examples of Some High Voltage Power Supplies
Reference Voltage Source
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 2 [7.C.2], Six bit to 32-step logarithmic staircase generator.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-19
-10V
High Voltage Power Supply Design Guide for Space
*G = voltage gain LOW GAIN AMPLIFIER
Figure 3 [7.C.3]. Signal processor.
V p
Figure 4 [7.C.4], Signal processor output waveforms.
7-20
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 5 [7.C.5]. Oscillator, modulator, and parallel charge multipliers.
POSITIVE DC SHUNT
NEGATIVE DC SHUNT
Figure 6 [7.C.6]. High voltage DC shunt regulators.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-21
High Voltage Power Supply Design Guide for Space
Figure 7 [7.C.7], Multiplier, low level, and regulated outputs.
REFERENCES
1 . Doong. H., and M. Acuna. “A High-Voltage Stepping Supply with Rapid Settling Time.” NASA/GSFC Publication
X-690-74-65, NASA Goddard Space Flight Center, Greenbelt, Maryland, March 1974.
2. Peletier, D.P., “A High Performance 4500 Volt Electron Multiplier Bias Supply for Satellite Use,” presented to
IEEE Nuclear Sci. Symp., Miami, Florida, December 5, 1972.
3. R. Greenburg. Motorola Power Transistor Handbook, Chapter VI
End of Document by H. Doong.
7-22
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.C.8. HAPI-PAPI programmable HV supply.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-23
HAPI-LAPI PROGRAMMABLE HV SUPPLY
High Voltage Power Supply Design Guide for Space
HV GENERATOR
MODULATOR
STAIRCASE
GENERATOR
DC-DC CONVERTOR
Figure 7.C.9. Identifying the circuit boards in Figure 7.C.8.
7-24
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.C.10. HV shunt regulator, enlarged.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-25
High Voltage Power Supply Design Guide for Space
Comments on HAPI L API:
In certain spacecraft, such as HAPI LAPI, DE and ISEE, the particle detectors need stepped high voltage power sup-
plies to rapidly change the DC voltage on the detecting plates over 3 or 4 orders of magnitudes. This is done many
times during each orbit.
In the HAPI LAPI supply presented here the speedy stepping occurs on the decrease of V out . In other words, it takes
20 ms to reach maximum V out , and then decreases at the faster rate of 300 V/m s .
In Figs. 1 and 6 [Figs. 7.C.1 and 7.C.6, respectively], as shown on the previous pages, the high voltage DC shunt
regulators are actually 9 transistors each, to obtain a 9 x 300 V or 2700 V capability each. The output load is about
50 to 100 pF capacitance.
To get a wide V out range, the lower voltage steps are derived from +LV and -LV low voltage supplies while the high
voltage is turned off. This occurs between V out equal to 0.3 to 10 V.
To ensure the high speed of downward stepping, the HV DC shunt regulators were not potted, but only conformal
coated and staked. The dielectric constant k e of the potting would reduce the speed of stepping by a factor of k e . Also
the k e distorts the wave shapes. The coating was applied by spraying, not brushing. The parts were spot-staked.
This supply had no capability of changing polarity. It also was quite large in physical size.
By contrast, the ISTP stepping supply of 1993, has to go + and - on each output; also the output voltage has to be
±5 kV on each output, or twice that of HAPI LAPI. Furthermore, ISTP has to step fast on increasing voltage, as well
as decreasing, at 500 V/ms; therefore, a new version of shunt regulator is being used, namely Optocouplers, instead
of varying conductivity of transistors (Fig. 7.C.11).
EQUIV.
EQUIV.
Figure 7.C.11. Basic optocoupler concept.
7-26
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
The Optocoupler uses infrared diodes to shine light. The reverse bias leakage current of a second set of high volt-
age diodes is enhanced when infrared light is shed on them. High voltage diodes such as 15 kV diodes of sufficient
peak inverse voltage (PIV) rating should be used. Diodes of the worst switching times are best for this application
(developed in Germany, and used by Jim Van Diver at the Univ. of New Hampshire). The LEDs were bonded to the
HV diodes with Uralane 5753 and the resin acted as a light pipe and enhanced the effect, rather than weakened it by
light absorption. Of course, the LEDs require power, that is 200 mA peak at 3 or 4 V.
Whereas originally, technical information on optocouplers products was obtainable from Hewlett Packard Co., at this
time, Micropac Industries, Inc. is one of the suppliers of Optocouplers.
Again, just as in HAPI LAPI, the supply must only be coated, to maintain speed of voltage change. At a possible
voltage difference here of 10 kV maximum, this presents problems. Meticulous cleanliness and excellent adhesion
is required.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-27
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE D
-30 kV DC Power Supply for the Charge, Energy, and Mass (CHEM) experiment on the Active Magnetospheric
Particle Tracer Explorer (AMPTE, also called “Ulysses”) mission. Space Physics Group, Department of Physics,
Univ. of Maryland, College Park, Maryland 20742.
An interesting -30 kV DC power supply was built by the University of Maryland for the CHEM experiment on the
AMPTE project. It had several AC voltages riding on the -30 kV DC, such as ±1 kV AC pk-pk. This supply was almost
bare, not potted and not conformal coated, with exception of one transformer discussed further on. The high vacuum
of space was relied upon as insulator during the mission, and in order to obtain and maintain this high vacuum within
the shielding box, these things were done: (1) one month of outgassing before turn-on in orbit; (2) large, screened-over
vent hole in the shielding box; and (3) use of low vapor pressure parts and materials — avoidance, as far as possible,
of polymer coatings.
In Fig. 7.D.1 (photo) of five figures, the transformer shown on the right-hand side, on the underside of the board, is the
pot-core transformer (output of 1000-1500 V, pk-pk) in a cylindrical aluminum shielding can. This output gets multi-
plied by the HV multiplier stack. In Figs. 7.D.1 and 7.D.4, the large transformer on the left-hand side is the transformer
for the “smaller” AC voltages. This transformer is also at -30 kV DC with respect to “ground,” and therefore, has to
be of physically larger dimensions to try to prevent corona to ground. In the background of Fig. 7.D.1, the cylindrical
insulating liner can be seen, containing the longitudinal rails upon which the entire circuit board rides. The metallic
cylindrical shield goes around it. On several of the photos, the “field smoothers” or “hoops” are seen. Another term
for these could be corona rings or corona doughnuts.
Figure 7.D.2 is a closer view of the HV multiplier stack. Note that the multilayer ceramic capacitors (Johansen) are
not coated and have no lead wires. They are soldered to rectangular short trails on the G-30 circuit board, 20 mils
up from the board with SM 63 solder at 600°F. In other words, before soldering, a 20 mil shim is slipped under the
capacitor lying flat, to be later removed. The capacitors thus ride on 20 mil high stilts of solder which serve as a
thermomechanical stress relief between the capacitor’s ceramic and the polyimide circuit board. Also, the 20 mil gap
helps the outgassing. The diodes in the HV stack were Semtech SS4710.
Figures 7.D.3 and 7.D.4 show the other side of the circuit board. In Fig. 7.D.3, the resistors are seen to be mounted up,
off the board; they are Victoreen MOX type. They are spot-bonded in the middle to the board, with very low vapor
pressure material, Chemglaze Z004 (or could be Epon 934).
Figure 7.D.4 shows the circular High Voltage transformer on the left, having several high voltage outputs. The
transformer wire was bought from Phelps-Dodge with Polythermaleze 2000 magnet wire insulation. The particular
transformer shown here was in addition coated with a Parylene coating of 1 or 2 mil thickness. This coating was not
successful on the particular serial number shown here; it is seen to have bad adhesion and is peeling. The Parylene later
was made to adhere in other serial numbers. On the right-hand side of Fig. 7.D.4, one can see, around the threaded
golden screw hole, the pads for the pins of the output connector to ride on. This is a custom designed, special connec-
tor by the Univ. of Maryland. Figure 7.D.5 shows an outside view of it.
Cleaning of the circuit board was done with hot ethyl alcohol, but not the transformers. All screws/screw holes were
vented to prevent corona from forming in the dead-end holes under the screws. This power supply was tested for 3
or 4 weeks, continuously in high vacuum, with thermal cycling from -45°C to + 45°C. In summary, this type of bare
construction on a -30 kV DC power supply requires great care in handling, meticulous cleanliness, and long vacuum
testing.
Nearly identical versions of the -30 kV DC power supply described above were flown as integral portions of time-of-
flight (TOF) ion composition spectrometers designed to measure plasma composition and energy spectra in various
7-28
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
space environments. These TOF instruments were flown successfully on AMPTE (launched August 1984), Ulysses
(launched October 1990), Wind (launched November 1994), and ACE (launched August 1997). After an outgassing
period of four to six weeks following launch, the -30 kV supplies were slowly ramped up to operating voltage over a
period of about five to seven days. Operation voltage was typically -23 to -28 kV. The AMPTE instrument (Gloeckler
et al., 1985) and its -30 kV supply operated flawlessly until the end of mission, about four years after launch. The
-30 kV supply of the TOF instrument (Gloeckler et ah, 1995) on the Wind Spacecraft was of a slightly modified design.
It operated initially at -23 kV and later at -28 kV until May 27, 2000 when a reset of the instrument Data Processing
Unit occurred and the instrument ceased to operate. The Ulysses (Gloeckler et ah, 1992) and ACE (Gloeckler et ah,
1998) TOF instruments (the Solar Wind Ion Composition Spectrometers, or SWICSs) have been operating flawlessly
since turn on. Currently, the Ulysses instrument is at -24 kV and the ACE at -26 kV. To the best of our knowledge,
none of the -30 kV supplies described here and developed and flown by the Maryland Space Physics Group have
failed. Together, these -30 kV supplies have, as of April 2005, a combined flawless record of operation of 31 years
and 9 months.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-29
High Voltage Power Supply Design Guide for Space
Figure 7.D.1
7-30
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.D.2
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-31
High Voltage Power Supply Design Guide for Space
Figure 7.D.3
7-32
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.D.4
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-33
High Voltage Power Supply Design Guide for Space
Figure 7.D.5
7-34
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
REFERENCES
Gloeckler, G., F.M. Ipavich, W. Stuedemann, B. Wilken, D.C. Hamilton, G. Kremser, D. Hovestadt, F. Gliem, R.A.
Lundgren, W. Rieck, E.O. Turns, J.C. Cain, L.S. MaSung, W. Weiss, and P. Winterhof, “The Charge-Energy-Mass
(CHEM) Spectrometer for 0.3 to 300 keV/e Ions on the AMPTE CCE.” IEEE Trans. Geosci. Remote Sens., GE-23(3),
234-240, 1985.
Gloeckler, G., J. Geiss, H. Balsiger, P. Bedini, J.C. Cain, J. Fischer, L.A. Fisk, A.B. Galvin, F. Gliem, D.C. Hamilton,
J.V. Hollweg, F.M. Ipavich, R. Joss, S. Livi, R. Lundgren, U. Mall, J.F. McKenzie, K.W. Ogilvie, F. Ottens, W. Rieck,
E.O. Turns, R. von Steiger, W. Weiss, and B. Wilken, “The Solar Wind Ion Composition Spectrometer,” Astron. As-
trophys. Suppl. Ser., 92, 267-289, 1992.
Gloeckler, G., H. Balsiger, P. Bochsler, A. Buergi, A.B. Galvin, J. Geiss, F. Gliem, D.C. Hamilton, T. Holzer, D.
Hovestadt, F.M. Ipavich, E. Kirsch, R. Lundgren, K.W. Ogilvie, and B. Wilken, “The Solar Wind and Suprathermal
Ion Composition Experiment (SMS) on the Wind Spacecraft,” Space Sci. Rev., 71, 79-124, 1995.
Gloeckler G., P. Bedini, P. Bochsler, L A. Fisk, J. Geiss, F.M. Ipavich, J. Cain, J. Fischer, R. Kallenbach, J. Miller, E.O.
Turns, R. Wirnmer, T. Zurbuchen, “Investigation of the composition of solar and interstellar matter using solar wind and
pickup ion measurements with SWICS and SWIMS on the ACE spacecraft,” Space Sci. Rev., 86, 495-537, 1998.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-35
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE E
A commercial, -700 V power supply. BBXRT Project, Short Mission, by, Renate S. Bever, NASA/GSFC.
If the mission duration lasts only a short time, such as one week on a shuttle flight, or a rocket flight, some project
offices have recommended certain commercial small power supplies. They are already internally solid-potted and
can then be modified either at the output or at the input terminals to meet required specifications.
In the lower half of Figs. 7.E.2 and 7.E.3 is a Velonex Company solid-potted power supply, variable from -300 to
-700 V DC by means of a small accessible variac screw at the top of the housing. Next to the Velonex supply sits a
circuit board at the output end, with high voltage capacitors and resistors to serve two purposes: (1) slow application
of voltage to the instrument, about 20 s rise-time, and (2) filtering. All this is mounted in a rectangular shielding box
approximate size 6.5 in x 2.5 in, to be primed and solid-potted.
Several new experiences were gained here:
(1) It had to be roughed on the five surfaces (not the top) with 400 grade emery paper and well-primed with PR 420
before potting with Conathane EN-11, all to enhance adhesion.
(2) Thermal analysis had shown that EN-11 did not have sufficient thermal conductivity to remove the heat from the
Velonex supply buried in the potting compound. Therefore, powdered Cho-therm was added to some EN-11, and
the Velonex supply pasted with a thin layer of this mix to the bottom of the shielding box before the general pot-
ting. This then enhanced good thermal conduction to the outside shielding box.
(3) The output cable from the Velonex supply was polyethylene, which does not adhere well to EN-11. It had to be
roughed up and primed, as well as the other circuit components primed before potting by vacuum pouring with
EN-11, just below the level of the top of the Velonex supply.
(4) After the Project office decided on the output voltage, (first specified as -500 V, but after one year determined to
have to be -700 V), the variable screw at the top was fixed with a few drops of EN-11.
(5) The output was by means of a female Reynolds series 600 connector on a cable pigtail. The male Reynolds bulkhead
connector is much more difficult for obtaining good adhesion.
7-36
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
cc
o
I —
o
LU
I —
LU
Q
O
Figure 7.E.I. Velonex supply; filtering and grounding externally.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-37
High Voltage Power Supply Design Guide for Space
Figure 7.E.2. BBXRT supply before potting. Figure 7.E.3. BBXRT supply after potting.
7-38
Chapter 7: Design Examples of Some High Voltage Power Supplies
OVERALL LENGTH
225 CM
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
DESIGN EXAMPLE F
A 3-5 kV, also 1.3 kV High Voltage power supply for the EGRET/GRO project. Design Engineer: Arthur Ruitberg,
GSFC, Code 563.
OVERALL DIAMETER
165 CM
LIGHT SHIELD
HERMETICALLY SEALED
ELECTRONICS BOX /HV SUPPLY
ANTICOINCIDENCE
SCINTILLATOR DOME
UPPER SPARK CHAMBER
ASSEMBLY
UPPER SCINTILLATOR
LIGHT PIPE AND PM TUBE
ASSEMBLY
LOWER SPARK CHAMBER
ASSEMBLY
ANTICOINCIDENCE
PM TUBES
PRESSURE SHELL
ASSEMBLY
LOWER SCINTILLATOR
LIGHT PIPE AND PM TUBE
ASSEMBLY
BULKHEAD PEDESTAL
ELECTRONICS BOXES
GAS REPLENISHMENT
SYSTEM
NalPM TUBES
Nal CRYSTAL
Figure 7.F.I. EGRET Spark Chamber (PM=PhotoMultiplier).
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-39
High Voltage Power Supply Design Guide for Space
The EGRET Spark Chamber high voltage power supply has really two high voltage supplies in it, as can be seen on
the circuit schematic:
(1) The Spark Chamber Supply giving between 3-5 kV output and taking over 50 ms to restore the output voltage
after a spark.
(2) A separate 1300 volt generator: The Spark Chamber supply works on a fly-back transformer design. The fly-back
transformer is T[ with a 1:10 turns ratio. The controlling amount of energy is in T,; when transistor Q 3 is turned off,
the negative current in the primary of Ti continues. The low voltage portion is a Jensen oscillator with transformer
T 3 controlling the frequency.
The high voltage portions of this power supply were solid potted with EN-11 polyurethane as seen in the Figure 7.F.8
photograph, below the “cut end” of the photo. This was in addition to the fact that the entire high voltage power supply
for the EGRET Spark Chamber was pressurized at more than atmospheric pressure (probably with dry nitrogen gas).
In case pressurization would be lost, that way the power supply would still function, even when the pressure would
drop through the dangerous “corona region.”
Please note that Figure 7.F.2 actually has two power supplies on it: the bottom, righthand part of the page is the 1.3
kV supply. The other three-fourths of the page is the 3-5 kV supply (originally to be 2-4 kV).
7-40
Chapter 7: Design Examples of Some High Voltage Power Supplies
<!» 2-5 KV
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.F.2. EGRET Spark Chamber high voltage power supply schematic diagram.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-41
EGRET SPARK CHAMBER HIGH VOLTAGE
POWER SUPPLY SCHEMATIC DIAGRAM
High Voltage Power Supply Design Guide for Space
Controlling amount of energy is in Ti; when Q3 is turned off, then the negative current in the
primary continues.
Figure 7.F.3. Power transformer parasitic capacitance is the “heart” of the circuit as in Figure 7.F.2.
7-42
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.F.4. Control loop; detail of Figure 7.F.2.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-43
R12, 1%
RANGE ADJ
2- 3 KV, R=1 .91 K, 1 %
2. 5-3. 5 KV, R=8.06 K. 1%
3- 4 KV, R= 16.9 K, 1%
High Voltage Power Supply Design Guide for Space
Vin b R17IIR18
Vp * 1+ SC (R34 + R 17 ||Rib)
Vin R 17 IIR 18
RESPONSE 4/18/86
1
i
1
i
SiCrV
r
•
»
RESPONSE
Over 50 milliseconds to restore the output voltages
Figure 7.F.5. Integrator Performance versus Ability to Calculate.
7-44
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.F.6. FIV layout, i.e., top of Figure 7.F.7.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-45
High Voltage Power Supply Design Guide for Space
Figure 7.F.7. Identifying major parts of the layout.
7-46
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.F.8. Photograph of EGRET FIV supply.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-47
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE G
XTE/PCA High Voltage Power Supply (really two power supplies to about 2200 V and 2900 V, Design Engineer:
Karen Castell-Stewart, NASA/GSFC
FEATURES
• Two independent, isolated high voltage power supplies packaged in one box.
• Each output is programmable (with 4 bits) to 16 voltage levels.
• An additional bit command the output on or off.
• Xenon Output: 980 V unregulated; 1910 V-2190 V regulated, 20 V steps.
• Propane Output: 1430 V unregulated; 2580 V-2860 V regulated, 20 V steps.
• Output Voltage Regulation: ±0.2% over temperature and input voltage variations.
• Output Ripple and Noise: <lm V p-p over temperature and input voltage variations.
• Operating Temperature Range: -10°C to ±40°C.
• Isolation: The input and output power returns shall be isolated from each other and the chassis.
• Transient Response: <4% overshoot to worst case step command with a time duration of <400 ms.
• Slow Turn On: Upon turn on, the output shall rise in >100 ms to prevent damage to the detectors.
• Control Loop: In the event that the control loop opens, a second, passive loop shall prevent the output voltage
from exceeding the maximum step voltage.
• Output Voltage Monitoring: A telemetry readout shall provide a -5 V to ±5 V output, proportional to the high
voltage; that applies to the upper 8 and lower 8 settings independently.
• Short Circuit Protection: The supplies must limit the output current in the case of a short applied to the output.
This spacecraft flies through the South Atlantic Anomaly (SAA) again and again.
In this region over the Southern Atlantic Ocean, the Earth’s magnetic field is abnormally large, and thus, there is an
abnormal concentration of high energy particles or a radiation belt. To protect the detectors from damage, the high
voltage needs to be turned down, hence the detectors off, on every passage through the SAA.
GENERAL DESCRIPTION
The XTE/PCA instrument requires two high voltage sources to bias the wire grids of its x-ray detectors. There is one
High Voltage Power Supply (HVPS) box to supply these voltages on each power control unit (PCU). It is located on
the PCU bottom cover, in electronic box stack #1. Each HVPS box consists of two high voltage supplies, one for the
xenon and one for the propane chamber. The supplies convert the filtered ±28 V DC spacecraft bus voltage, from the
Remote Interface-Power Supply (RIF-PS), to the regulated, high voltage DC needed for the detectors.
Each supply is individually programmable from a 5-bit command interface. This consists of single bit enable/disable
signal and an additional 4 bits that command 16 different output voltage levels. The upper 15 levels (0001-1111) are
20 V apart and the lowest step (0000) is about 1000 V below step (0001). Specifically, the xenon side is capable of
producing a 1910-2190 V regulated output, along with the lowest level, a 910 V unregulated output. Similarly, the
propane output ranges from 2580-2860 V for the upper 15 levels and 1400 V for the lowest level. This bottom step,
called the South Atlantic Anomaly (SAA) response, is intended to protect the detectors from damage as it flies through
this radiation belt by lowering the supplied high voltages by about 1000 V. Because the detectors will not be taking
data during this time, unregulated power was sufficient.
Another safety feature of the HVPS is the enable/disable bit. It will turn off the high voltage output of the power
supply with power still applied to the input. This serves to protect the detector against possible damage from the
7-48
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
high voltage without needing to turn off the power at the HVPS input. (The input power can be turned off as well by
signaling a relay on the RIF-PS board. Thus, the high voltage output can be turned off in two different ways and can
be turned down with the SAA signal.)
The output enable/disable circuitry is located at the front end of the power supply. As stated above, this is commanded
externally by one bit in the command interface. When its level is low, the high voltage supply operates normally.
When the level goes high, an optocoupler output shorts a zener diode in the preregulator, bringing the output down
to zero. Thus, the high voltage output falls to zero with input power still applied. An optocoupler is used in order to
maintain primary to secondary isolation in the PCU.
The board layout and component assembly for both the xenon and propane units was designed to be identical, to in-
crease the efficiency in schedule and budget. Some of the component values differ in order to accomplish the different
voltages required. Each assembly consists of a printed circuit board (PCB) and a high voltage assembly, sometimes
called the “bathtub.” The PCB is a two-sided board on which the low voltage and control circuitry is mounted. The
bathtub unit houses all of the high voltage components, except for the high voltage transformer, which is mounted
on the PCB. The bathtub compartment is encapsulated with CONAP EN-1 1 to reduce the risk of partial discharge in
areas of high electric field gradient.
CIRCUIT OPERATION
As described, the inputs to each high voltage supply are the +28 V line and 5 digital command bits (command inter-
face) to set the high voltage level. As the +28 V line enters the 1TVPS, it sees an inrush current limiter, a common
mode choke and an input filter. This filtered line then passes through a pre -regulator that brings the voltage down to
20 V DC. This DC voltage appears at the front end of Elartley oscillator, whose output is a 40 V p-p sinusoid. The
sinusoidal voltage is seen across the primary winding of a step-up transformer, which increases the AC voltage to
about 600 V p-p to 80 V p-p, depending on whether the supply is a xenon or propane.
On the secondary side, the stepped-up sinusoid drives the high voltage section of the power supply. The high voltage
is entirely contained on the secondary side of the transformer. Regulation is accomplished by placing two regulating
transistors in series with the transformer secondary and the voltage multiplier. In effect, the voltage at the input of
the voltage multiplier is adjusted by the regulating transistors. This then determines how much voltage appears at
the output. The voltage multiplier is a 4-stage Cockcroft-Walton multiplier, so its output is approximately 4 times
the voltage seen at its input. An output filter follows the multiplier and filters this DC high voltage in two stages in
order to achieve low noise output. At this point, a high voltage connector supplies the output to the detector through
cabling designed for high voltage use.
The high voltage output is divided down within the supply to provide a feedback signal to the feedback and control
circuitry. Several operational amplifiers use this error signal, along with a multiplexed signal from the 4 command
bits to set the level of the high voltage output. These signals can independently change the conduction of the regulat-
ing transistors to adjust the output up or down.
Another feature of the supply is an output voltage monitor which provides a low voltage level output corresponding to
the high voltage output. The monitor output is restricted to the range -5 V to +5 V with at least 0.5 V needed between
output steps. In actuality, the upper eight levels range from -0.5 V to +5.0 V with 0.7 V between output steps. The
lower seven levels are also in this range with the same increment between steps, but the lowest level, the SAA, is about
-4.5 V, a value far below the other steps, in order to easily distinguish this output condition from the others.
PACKAGING CONCEPT
The HVPS packaging concept was designed with a modular approach, such that all the high voltage components are
contained in a separate, removable unit. This high voltage container, also called the “bathtub” because of its shape,
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-49
High Voltage Power Supply Design Guide for Space
is installed on top of the PCB with the other, low voltage circuitry, and can be replaced with another bathtub unit if
a failure occurs. Thus, rather than having to replace the entire power supply in the case of an integrated design, the
modular approach saves most of the supply while allowing replacement of only the high voltage assembly.
The high voltage unit is divided into several compartments. This is to reduce noise coupling between sections and
to the output. The first chamber is the voltage multiplier section, which houses a 4-stage Cockcroft-Walton voltage
multiplier. These are Maida ceramic X7R capacitors, chosen for their successful space flight history. Additionally,
Renate Bever has done extensive testing of these parts. To reduce noise, the AC and DC capacitors where placed on
separate boards. The DC side is placed closer to the output in order to reduce cable length. The next two chambers
have output filters that reduce the high frequency AC noise. Two stages are needed to meet the ripple requirement of
<lmVp-p. A large series resistor in the second filter section provides short circuit protection in the event of a short
in the load. Because the transformer secondary winding is not in the high voltage section its HV winding had to be
wrapped in Kapton. The leads, where they come out, were coated and staked. Also, special attention had to be paid
to the parts in the HV Regulation portion of the circuit. The output connector is a Reynolds series 167-3771 connector
rated for 10 kV. The last section contains two Caddock high voltage resistors that divide down the high voltage for
the feedback control loop.
The entire high voltage enclosure is made of Noryl, which exhibits excellent insulating properties, but is very rigid,
and therefore, difficult to machine. It is also a lightweight material. The outside shielding is done with copper tape
with conductive adhesive. (Experiments were done with depositing aluminum coating on the Noryl, but the long
term adhesion was not good.) The shields separating the compartments are made of copper-clad epoxy, which are all
connected to signal ground.
The weight budget on XTE/PCA allowed for the encapsulation of the high voltage assembly. The encapsulant chosen
was CONAP EN-11, based on recommendations by Art Ruitberg and Renate Bever.
7-50
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-51
NOTE: RESISTOR VALUES SHOWN ARE FOR XENON | PROPANE| FLIGHT UNITS.
High Voltage Power Supply Design Guide for Space
Figure 7.G.2. High voltage before potting (lower half of figure).
7-52
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.G.3. High voltage after potting (lower half of figure).
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-53
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE H
Cassini/CAPS: -16 kV and +16 kV High Voltage Power Supply. Modular Construction. Design Engineer: Arthur
Ruitberg, NASA/GSFC
REQUIREMENT SUMMARY
• Two outputs, linearly controlled by separate eight bit command:
(1) 0 to +16 kV
(2) 0 to -16 kV
• DC regulation: 1%, ripple: <.01%
• Third output for mcp controlled +1.2 kV above negative output
• Operates from +30 V s/c bus, oscillators lock to 100 kHz input sync signal, 2.25 W available
• Qualification temperature range: -35 to +45°C
• Size: 8” x 6” x 1.9”, Weight: 1.24 kg
• Radiation level: 100 krad
The Cassini/CAPS (Cassini Plasma Spectrometer) power supply is, in very short form, included here. It has a -16 kV,
and a +16 kV, and also a +1.2 kV above the negative voltage, outputs. As the photograph shows, it is of modular con-
struction. That is, even within the high voltage section, it is neither all solid potted nor all just conformal coated. The
two filter sections at the left side of the photograph are potted with EN-11 polyurethane as rectangular “blocks.”*
The two high voltage stacks have the capacitors and diodes placed or “stuffed” within premachined housings. The
transformers at the right-hand side are contained in molds made of Ultem 1000 by GE. Connections between these
circuit elements are made with smooth surface, bare metal rods. Junctions of these rods are made within smooth, small
metal corona balls, rather than soldered. Where these rods penetrate shielding box walls, they go through porcelain
smooth surface insulators, etc. The entire voltage supply was assembled, certain areas were masked (see Appendix
II of Ch. 3), and then the assembly was Parylene-coated, also according to Appendix II of Ch. 3. The reason that this
packaging technique could be done for a 16 kV supply was that during the seven year flight to the vicinity of planet
Saturn, the inside of the power supply would have been thoroughly outgassed, and the pressure within the supply
would then be deep-space vacuum, an excellent insulator. The obvious advantages were the saving of weight, the ease
of removal of a failed component during construction/testing, and also a considerable reduction of thermomechanical
stresses during thermal excursions, which always exist within a large potted volume. At the time of this writing, the
HV supply has already been operating well for five or six months.
* Instead of EN-11A/B, it was EN-4A used with EN-11B. EN-11A is no longer available. See Chapter 4, Appendix II.
7-54
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.H.I. Cassini/CAPS HVU-1.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-55
High Voltage Power Supply Design Guide for Space
7-56
Chapter 7: Design Examples of Some High Voltage Power Supplies
PSL14
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Cl C2 C3 C4 Q22
470pF, 5KV 470pF, 5KV 470pF, 5KV 470pF, 5KV 470pF, 5KV
.04X45°
(TYP)
1. MATERIAL: ULTEM 1000 (GE PLASTICS)
(POLYETHERIMIDE RESIN)
MATL WILL BE SUPPLIED BY CODE 692
2. REMOVE BURRS AND BREAK SHARP
EDGES
Figure 7.H.4
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-57
High Voltage Power Supply Design Guide for Space
DESIGN EXAMPLE I
Anti-Coincidence Detector (ACD) High Voltage Bias Supply for ACD/GLAST project. Up to 1,300 V DC
Design Engineer: Arthur Ruitberg, NASA/GSFC
1.1 High Voltage Bias Supply (HVBS) Specification for ACD/GLAST
1.1.1 There shall be 24 HVBS, one for each eight phototubes
1.1.2 The high voltage bias supply shall operate from a +28 V ±6 V power supply. The ripple specifica-
tion on this input power is < TBD mV rms.
1.1.3 The HVBS shall dissipate no more than 150 mW
1.1.4 The output shall be adjustable from 0-1300 V, ±5%
1.1.5 The output current shall limit at 40 pA
1.1.6 The ramp-up time constant shall be greater than 5 s and less than 30 s.
1.1.7 The ramp-down time constant shall be greater than 5 s and less than 30 s.
1.1.8 The mass shall be less than 100 g.
1.1.9 The overall volume of a single HVBS shall not exceed TBD mm x TBD mm x TBD mm.
1.1.10 The temperature stability of the HVBS shall be <5000 ppm/°C
1.1.11 The temporal stability of the supply shall be <0.5%/day
1.1.12 The HVBS shall provide an output monitor of 0-2.5 V that is proportional to the HV output to
within 1%.
1.1.13 The HVBS oscillator frequency shall be >100 kHz
1.1.14 The output ripple/noise shall not exceed 2 mV peak-peak from 50 Hz to 50 MHz
1.1.15 The HV return shall be isolated from both chassis andHV supply ground by 100 Q ±20 Q. The HV
return shall be grounded to the analog ground at the PMT with a DC resistance of < 100 mQ.
1.1.16 All flight electronics surfaces shall be conformally coated with a NASA-approved polymer.
1.1.17 The HVBS will be enclosed in a conductive housing to provide EMI/EMC shielding. The enclosure
surfaces will be plated with a non-oxidizing conductor.
1.1.18 A 1.5 mm ±0.5 mm diameter vent hole will be provided in the housing to ensure rapid venting.
Abbreviations to make subsequent pages readable:
ACD Anti-Coincidence Detector
BEA Base Electronics Assembly
BFA Base Frame Assembly
DAC Digital to Analogue Converter
EC Electronic Chassis
FREE Front End Event Electronics
GAFE GLAST ACD Front-end Electronics
GARC GLAST ACD Readout Chip
GLAST Gamma-ray Large Area Space Telescope
HVBS High Voltage Bias Supply
LAT Large Area Telescope
PCB Printed Circuit Board
PCU Power Control Unit
PMT Photo-Multiplier Tube
RN Resistor Network
SAA South Atlantic Anomaly
7-58
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
TDA Tile Detector Assembly
TSA Tile Shell Assembly
Packaging Considerations
Since the maximum voltage on anyone of these power supplies is only 300 V, then conformal coating with Parylene
by vacuum vapor deposition was used. A process specification for Parylene coating is included in this book as an
Appendix to Chapter 3. Although that particular specification is for Cassini/CAPS portions, the basic process is the
same, although the parts to be “masked” are different.
General Comments
The ACD/GLAST High Voltage Bias Supply (HVBS) for the Photomultiplier Tubes (PMT) is an example of Mass Pro-
duction. As seen on the Specification page: “There shall be 24 HVBSs, one for each of eight phototubes.” In other words,
there are 24 HVBSs and 24x8 or 192 high voltage distribution boxes from the 24 power supplies to the phototubes.
This then is the reason for using the relatively new, in high voltage work, Flex Circuit Board. This has three layers of
insulation adhered together, namely 0.002-in Kapton Flex between three layers of rigid Polyimide, for finished rigid
thickness of about 0.062 in; and 4 layers of metallization trails and pads: Namely one layer on each side of the inner
Kapton Flex and one layer on the outside of each of the Polyimide layers. These metallizations are deposited “to order”
by the manufacturer. See the accompanying drawing and the color print of the Corona Test Object. The two 0.002 -in
thick “ribbons” of Kapton Flex connecting the three small square circuit boards have straight metallization trails on
them, thus making electrical connections between proper points on the small square circuit boards. The Kapton “rib-
bons” are flexible and can be bent into U shapes or semicircular arcs. This enables several of the small square circuit
boards to be stacked above each other into a relatively small shielding box, so some of the tedious assembly work for
the large number of units in ACD/GLAST is done with relative ease.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-59
FABRICATION MOTES
High Voltage Power Supply Design Guide for Space
Figure 7.1.1.
7-60
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
GLAST Photomultiplier Flex Board PWB 2054578 Rev - 02/9/04
Primary Component Side
Layer 1 of 4
Flex Layer 2
Layer 2 of 4
Flex Layer 3 Secondary Component Side
Layer 3 of 4 Layer 4 of 4
Figure 7.1.2.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-61
High Voltage Power Supply Design Guide for Space
7-62
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
CD
CD
5
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-63
15) FROM CHASSIS TO HARDWARE
High Voltage Power Supply Design Guide for Space
Figure 7.1.5. Photograph of HVBS.
7-64
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Figure 7.1.6. Photograph of some of the PMTs.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-65
High Voltage Power Supply Design Guide for Space
Appendix to Design Example I (HVBS)
Operational Guide for the GLAST ACD High Voltage Bias Supplies (HVBS)
DAS, 5/7/04
Purpose
It is possible that the intended operation of the ACD High Voltage Bias Supplies (HVBS) and their operational con-
straints, including the ways these work with the BEA electronics, is not immediately intuitive. This memo seeks to
document some of the design intent and explain the expected operation functions, including environmental pressure-
related instrument safety issues.
Table 7.1.1. Commands associated with the ACD HVBS.
Command
Function
Code
Register
Number
Data
Field
Comment
HV Level Nominal
0
8
12 bits
This is a register in the GARC
that contains the 12 bit value that
will be sent to the DAC with the
Use HV Nominal command.
SAA_HV_Level
0
9
12 bits
This is a register in the GARC
that contains the 12 bit value
that will be sent to the DAC with
the Use HV SAA command.
Use HV Nominal
0
10
N/A
This command transfes the
12 bits in the HV Level Nominal
register from the GARC to the
MAX5121 DAC.
Use HV-SAA
0
11
N/A
This command transfers the
12 bits in the SAA HV Level
register from the GARC to the
MAX5121 DAC.
GARC Mode
2
8
12 bit register, [11:0]
bits [3:1] are HV 1 Enables
bits [6:4] are HV 2 Enables
These are the command bits for
the Triple Modular Redundant
command bits for the two supplies.
GARCStatus
2
9
N/A
This is a read-only register.
Bit 1 is the HVBS 1 Enable status
Bit 2 is the HVBS 2 Enable status
7-66
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
The block diagram below may assist operators in conceptualizing the HVBS control functions.
HV Monitor 1 to GASU
HV Monitor 1 to GASU
Figure 7.1.7. Simplified block diagram of ACD BEA high voltage control
Power
The ACD HVBS operate on +28 V power supplied to the BEA through the 79 pin circular connectors. The pin locations
are identical for each of these connectors. There are two HVBS per chassis assembly. The BEA receives a nominal
+28 V on pins 5 and 7 with the +28 V _RETURN on pins 33 and 34. The HVBS cannot supply a high voltage when this
power supply is off (i.e., at 0 V). The FREE card uses a separate 3.3 V power supply and can operate independently from
the HVBS. It is important that the +3.3 V FREE card power be ON when the HVBS +28 V power is ON. The GARC
ASIC on the FREE card must be on to ensure proper control of the HVBS enables and DAC level commands.
HVBS Control
Once the +28 V power is applied to the HVBS, the FREE board controls the output level of the supplies by means
of an analog level voltage and via a digital enable level. The digital enable level is active high, meaning that +3.3 V
is considered ON and 0 V is considered OFF. The analog level voltage is sent to the bias supplies differentially. This
is controlled on the FREE card via the output of the MAX5121 Digital-to-Analog converter (DAC). The FREE card
has an amplifier circuit which converts the single-ended 0-1.249 V DAC output to a differential voltage of 0-2.5 V
with a IAV pedestal offset. The full range of the differential voltage of 0-2.5 V with a 1.4 V pedestal offset. The full
range of the differential voltage (i.e., 2.5 V differential) corresponds to a full-scale range on the bias supplies of ap-
proximately 1300 V.
Use of Triple Modular Redundancy in the High Voltage Enable Bits
The GARC Mode command is used to enable and disable the two ACD HVBS in each chassis. Each HVBS is en-
abled/disabled independently. In the nominal operational mode, only one supply is enabled at a time.
It is considered important that the FREE electronics maintain control of the HVBS output voltages at all times.
The use of Triple Modular Redundancy (TMR) in the logical constructs of the GARC seeks to alleviate the condi-
tion that a single-event upset (SEU) could alter the state of the HV Enable flip-flop. This is done by implementing a
majority voting logic for these circuits. Therefore, proper use of the GARC_Mode command means that the ground
command should set bit [3:1] all to the same value. This is also true for bits [6:4], (The one exception to this would
be in Functional Tests where the operation of the TMR circuitry is being checked). These GARC_Mode bits are the
inputs to the TMR voting logic. The outputs of the TMR logic are the physical level signals (HV _ENABLE_1 and
HV ENABLE 2) that are passed from the GARC to the HVBS 1 and 2. The status of these levels is monitored via
the read-only GARC Status command, bits 1 and 2, respectively.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-67
High Voltage Power Supply Design Guide for Space
It is important that during non-operational times (e.g., periods of instrument operation when the phototubes are desired
to be OFF) that these levels are controlled to the “0” or OFF state. This is the only means to keep the HVBS outputs
at a zero level. If the DAC is at a “zero volts” level, but the HVBS is enabled, the flight supplies may still output over
100 V DC.
During flight operations, the GARCMode register values may be periodically checked to ensure no upset has occurred
or should be periodically reloaded with refreshed values. A rough estimate of the GARC upset rate would indicate
that once a week or once a month would be more than adequate in flight.
Operations within the South Atlantic Anomaly (SAA)
The GARC has been designed to account for proper phototube operation during SAA transitions. The SAA is a
region of charged particles pulled a bit closer to the Earth’s surface by a local increase in the Earth’s magnetic field.
The implication for the ACD, which is by definition a charged particle detector, is a very substantial increase in the
amount of light scintillating in the tiles in this region. Because of the radiation increase, no science observations will
be possible for the LAT during SAA transitions, which will occur approximately every 100 min (but will vary based
on the precession of the LAT orbit).
If the PMT high voltage is left at nominal levels during this time, there will be excess current flowing through the
tubes, resulting in a decreased lifetime of the detector system. It is desirable to decrease the PMT high voltage (which
is nominally about 1000 V for a gain of about 1 x 10 6 ) by several hundred volts to reduce the electron multiplication
gain by several orders of magnitude. However, it is also not desirable to turn of the HVBS completely due to the
mechanical stress associated with the cycling of high voltage capacitors. An intermediate approach is to maintain a
level of bias on the PMTs that it represents a very low electron multiplication gain while not allowing a substantial
electric field change within the high voltage capacitors. Therefore, it is more desirable to turn the PMT voltage down
to a number such as 400 V or 500 V. This is accomplished by having two levels stored in the GARC — a “nominal”
science operations level and a lower, protected level, which is designated as the “SAA level.” These values are stored
in registers in the GARC and sent to the DAC at the appropriate orbital times. \
One fact about the FREE electronics that is not necessarily intuitive is that the DAC output (which becomes the dif-
ferential analog level control for the HVBS) is controlled by a register in the MAX5121 DAC, not the GARC. The
values stored in the GARC are loaded into the MAX5121 register whenever one of the “Use Level” commands is
sent. This allows the ACD BEA operator to verify that the correct values arc in the GARC registers prior to enabling
transmission of the value from the GARC to the DAC.
One other advantage to this approach is refreshing the contents of the MAX5121 DAC control register. The MAX5121
is not a radiation-hardened part; it is a commercial device. A single-event upset (SEU) in the control register will be
reflected in the DAC output. The process of sending the “Use Nominal” and “Use SAA” commands provides an au-
tomatic memory-scrubbing cycle on this register, which is designed to mitigate the effects of SEUs on the DAC logic.
Additionally, operators should note that the value written into the MAX5121 DAC may be read backjust one time; this
part has the curious feature of a destructive read operation. This is the only GARC-addressable register which oper-
ates in this manner. This feature may be understood by doing a GARC Write command to one of the Use HV Level
Function/Registers and then doing a GARC Read command to the same Function/Register. This will return the value
from the MAX5121. Subsequent reads will produce a different data pattern; the read is destructive in nature. Reading
the MAX5121 does not affect the value latched into the part, however. Only the verification aspect is destructive.
Consideration of the Pressure of the Environment in which the HVBS Operates
The ACD HVBS are sensitive to the gas pressure under which they are operated. This is an ACD instrument safety
issue. Operators of the ACD must be conscious never to enable the high voltage bias supplies unless the pressure
7-68
Chapter 7: Design Examples of Some High Voltage Power Supplies
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
is approximately 1 atm (i.e., 760 torr nominal) or the pressure is less than 5 x 10' 6 torr in a vacuum environment. If
operated in a vacuum environment, the electronics should have been at the < 5 x 10' 6 torr limit for 12 hours or more
to ensure proper venting. If the HVBS are operated in between these regions (i.e., in the region 5 x 10" 6 torr < Pres-
sure < Ambient) , there is a possibility of a corona discharge occurring in either the HVBS or the PMT assemblies.
(We would expect arcing when the local pressure around a HV component is somewhere in the 10' 1 to 10 1 torr range).
This form of plasma discharge could cause permanent damage to the electronics of the BEA. There are two expected
environments where this is applicable: thermal vacuum testing and in-flight operations.
The responsibility for correct operation of the high voltage control must fall on the Test Conductor. It should not be
assumed that test scripts can be relied on to autonomously protect the instrument. Additionally, the concerns with
high voltage operations may not be obvious to those unfamiliar with the operation of the ACD.
Sample HVBS Power-Up Sequence
The following represents a strawman sequence of steps for correctly powering up a BEA HVBS. It is possible this
might be used to form a template that could be used for an actual script.
1. Verify that the ACD +3.3 V power is on and currents are nominal.
2. Verify that the BEA chassis is in the proper pressure environment (as described above).
3. Send an ACD GARCMode Read command. Verify bits [6:1] are 0.
4. Send an ACD GARCStatus Read command. Verify bits [2:1] are 0.
5. Verify that the ACD +28 V power is on and currents are nominal.
6. Verify the GASU HVBS Monitors 1 and 2 are reading ~0 V
7. Send an CD GARC Mode write command with bits [3:1] at b ill.
8. Send an ACD GARC Mode read command to verify this command.
9. Send an ACD GARC Status Read command. Verify bits [2:1] are b'Ol, indicating HVBS 1 is enabled and HVBS
2 is disabled.
10. Send the GARC HVBS Level Write command with a data argument of 300. Verify with a GARC Read com-
mand.
11. Send the Use_HV _Normal write command. Verify the DAC register contents with the Use_HV _Normal read
command.
12. Verify that the GASU HVBS Monitor 1 reads an equivalent of approximately 100 V and that HVBS Monitor 2
reads approximately 0 V. Verify that no phototube has a rate in excess of 1 kHz (if it does, stop at this point and
inform the science team).
13. Verify that the +28 V current is at a nominal value.
14. Repeat steps 10-13 with data arguments of 600, 900, 1200, etc., up to 3000. At a DAC register content of 3000,
the HVBS should be at about 1000 V. PMT rates should be less than 1 kHz.
Powering down the supply would be essentially the same sequence of steps, but in reverse (larger data argument steps
could be used, for example steps of 1000). The DAC register would be ramped down to 0 and then the GARC Mode
enable bits would be zeroed out.
Chapter 7: Design Examples of Some High Voltage Power Supplies
7-69
High Voltage Power Supply Design Guide for Space
AUTHOR’S NOTE
The chapters in this document have evolved over a period of several years and from Knowledge Cap-
ture meetings where exchange of experience and knowledge took place between several personnel. It
is hoped by the authors that engineers and technicians that are newly entering the field will find this
book helpful to them.
As seen in the last chapter — Chapter 7 — much work has been done at GSFC and elsewhere, on power
supplies that produce high voltage biasing for scientific instruments, but put out minimal current. De-
signs for Traveling Wave Tube (TWT) supplies or laser power supplies, which must give considerable
current as well as high voltage, are not included in Chapter 7.
Such high voltage- high current supplies are usually called Electronic Power Conditioners or EPCs.
These have several outputs with two or even more outputs delivering tens of watts to the TWTs (or
lasers). This means, for example, for a 60 W TWT, that 25 milliamperes (mA) — not microamperes
(pA) — are delivered by the EPC at 2500 V, by just a single output.
Obviously for these supplies, the low voltage portions of the circuits are very different and the parts
and the transformer windings are different, etc., than for any of the low current, high voltage supplies
discussed in this book. More feedback loops must be added for over-current and over-power protection.
Also, the mounting of parts and the encapsulating resin must be chosen to get the internally developed
heat (usually several watts) out of the supply to avoid thermal run-away.
Generally speaking, NASA relies on private industry to design and build these EPCs. In today’s world
of company buy-outs, finding names of manufacturers can be difficult. The authors of this book, have
supplied several names of such manufacturers in the United States — not to recommend one over
others — but to help the reader know where to begin to look, if necessary.
For TWTs:
Electron Technologies, Inc. (ETI), Divsion of L-3 Communications, Torrance, CA
Dynamic Wave Telecom, Inc., Anaheim, CA
Quarterwave Corp., Rohnert Park, CA
MCL, Inc., Bollingbrook, IL
There are also companies in Europe and Japan.
7-70
R.S. Bever, A.P. Ruitberg, C.W. Kellenbenz, and S.M. Irish
Authors ( Left to Right): Renate Bever, Arthur Ruitberg, and Carl Kellenbenz. (Sandra Irish was not
available the day the picture was taken.)
7-71
REPORT DOCUMENTATION PAGE
Form Approved
OMB No. 0704-0188
1
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing
data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or
any other aspect of this collection of information, including suggestions for reducing this burden, to Department of Defense, Washington Headquarters Services, Directorate
for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that
notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently
valid OMB control number.
PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YYYY) 2. REPORT TYPE
31-07-2006 Technical Publication
3. DATES COVERED (From - To)
4. TITLE AND SUBTITLE
High Voltage Power Supply Design Guide for Space
5a. CONTRACT NUMBER
5b. GRANT NUMBER
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S)
Renate S. Bever, Arthur Ruitberg, Carl W. Kellenbenz, and Sandra M. Irish
5d. PROJECT NUMBER
5e. TASK NUMBER
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES)
Goddard Space Flight Center
Greenbelt, MD 20771
8. PERFORMING ORGANIZATION
REPORT NUMBER
2005-02566-0
9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES)
National Aeronautics and Space Administration
Washington, DC 20546-0001
10. SPONSORING/MONITOR'S ACRONYM(S)
11. SPONSORING/MONITORING
REPORT NUMBER
TP-2006-214133
12. DISTRIBUTION/AVAILABILITY STATEMENT
Unclassified-Unlimited, Subject Category: 31,70
Report available from the NASA Center for Aerospace lnformation,7121 Standard Drive, Hanover, MD 21076. (301)621-0390
13. SUPPLEMENTARY NOTES
14. ABSTRACT
This book is written for newcomers to the topic of high voltage (HV) in space and is intended to replace an earlier (1970s)
out-of-print document. It discusses the designs, problems, and their solutions for HV, mostly direct current, electric power, or bias
supplies that are needed for space scientific instruments and devices, including stepping supplies. Output voltages up to 30 kV are
considered, but only very low output currents, on the order of microamperes. The book gives a brief review of the basic physics of
electrical insulation and breakdown problems, especially in gases. It recites details about embedment and coating of the supplies
with polymeric resins. Suggestions on HV circuit parts follow. Corona or partial discharge testing on the HV parts and assemblies
is discussed— both under AC and DC impressed test voltages. Electric field analysis by computer on an HV device is included in
considerable detail. Finally, there are many examples given of HV power supplies, complete with some of the circuit diagrams and
color photographs of the layouts.
15. SUBJECT TERMS
electric potential, high voltage, direct current, power supplies, bias, low currents, electrical insulation, coating, coating, resins,
circuits, electric corona, alternating current, circuit diagrams
16. SECURITY CLASSIFICATION OF:
17. LIMITATION OF
ABSTRACT
18. NUMBER
OF
19b. NAME OF RESPONSIBLE PERSON
Eric M. Young
a. REPORT
b. ABSTRACT
c. THIS PAGE
PAGES
Unclassified
19b. TELEPHONE NUMBER (Include area code)
Unclassified
Unclassified
Unclassified
230
(301)286-4822
Standard Form 298 (Rev. 8-98)
Prescribed by ANSI Std. Z39-18